|
22688
|
975
|
60
|
2026-05-12T07:17:44.977479+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570264977_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.575419,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.5865922,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5826017,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.6097366,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
8641226668622057640
|
-2543683444951237998
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22687
|
974
|
46
|
2026-05-12T07:17:44.960926+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570264960_m1.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
2003991523423783530
|
-2732257310247257465
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22686
|
975
|
59
|
2026-05-12T07:17:43.586623+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570263586_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Log In.
Log In.
[EMAIL]
Password
Login
Google
Forgot password?
Reset
Login via SAML SSO
Salaries, goals, expectations, technology…
We’d love to hear from you!
Take the survey
Take the survey
Did you know?
Did you know?
Planhat runs live training sessions
every
Tuesday
and
Thursday!
SIGN UP AT THE LINKS BELOW:
SIGN UP AT THE LINKS BELOW:
Tuesdays at 11am ET / 5pm CET
Tuesdays at 11am ET / 5pm CET
Thursdays at 10am ET / 4pm CET
Thursdays at 10am ET / 4pm CET...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.575419,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.5865922,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5826017,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.6097366,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Log In.","depth":9,"bounds":{"left":0.51662236,"top":0.3180367,"width":0.15292554,"height":0.06703911},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Log In.","depth":10,"bounds":{"left":0.51662236,"top":0.3140463,"width":0.088597074,"height":0.075418994},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"lukas.kovalik@jiminny.com","depth":9,"bounds":{"left":0.51662236,"top":0.40822026,"width":0.13297872,"height":0.04868316},"on_screen":true,"value":"lukas.kovalik@jiminny.com","help_text":"","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextField","text":"Password","depth":9,"bounds":{"left":0.51662236,"top":0.47685555,"width":0.13297872,"height":0.04868316},"on_screen":true,"help_text":"","role_description":"secure text field","subrole":"AXSecureTextField","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXButton","text":"Login","depth":10,"bounds":{"left":0.51662236,"top":0.5734238,"width":0.059840426,"height":0.04868316},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXLink","text":"Google","depth":9,"bounds":{"left":0.5831117,"top":0.5734238,"width":0.020279255,"height":0.04868316},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Forgot password?","depth":10,"bounds":{"left":0.51662236,"top":0.6472466,"width":0.044215426,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Reset","depth":11,"bounds":{"left":0.56083775,"top":0.6472466,"width":0.013630319,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Login via SAML SSO","depth":11,"bounds":{"left":0.51662236,"top":0.6775738,"width":0.04870346,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Salaries, goals, expectations, technology…","depth":11,"bounds":{"left":0.77726066,"top":0.44852355,"width":0.11353058,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"We’d love to hear from you!","depth":11,"bounds":{"left":0.7972075,"top":0.4708699,"width":0.07363697,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Take the survey","depth":10,"bounds":{"left":0.7642952,"top":0.5159617,"width":0.13962767,"height":0.04868316},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Take the survey","depth":11,"bounds":{"left":0.8121675,"top":0.5323224,"width":0.043716755,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Did you know?","depth":9,"bounds":{"left":0.76163566,"top":0.6300878,"width":0.14494681,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Did you know?","depth":10,"bounds":{"left":0.80701464,"top":0.6312849,"width":0.054022606,"height":0.023144454},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Planhat runs live training sessions","depth":10,"bounds":{"left":0.7875665,"top":0.66001594,"width":0.09291888,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"every","depth":10,"bounds":{"left":0.79238695,"top":0.6823623,"width":0.01662234,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Tuesday","depth":10,"bounds":{"left":0.8090093,"top":0.6823623,"width":0.024268618,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and","depth":10,"bounds":{"left":0.83327794,"top":0.6823623,"width":0.013464096,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thursday!","depth":10,"bounds":{"left":0.84674203,"top":0.6823623,"width":0.028922873,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"SIGN UP AT THE LINKS BELOW:","depth":9,"bounds":{"left":0.76163566,"top":0.73064643,"width":0.14494681,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SIGN UP AT THE LINKS BELOW:","depth":10,"bounds":{"left":0.80285907,"top":0.7302474,"width":0.062333778,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Tuesdays at 11am ET / 5pm CET","depth":10,"bounds":{"left":0.7947141,"top":0.74660814,"width":0.07862367,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Tuesdays at 11am ET / 5pm CET","depth":11,"bounds":{"left":0.7947141,"top":0.74660814,"width":0.07862367,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Thursdays at 10am ET / 4pm CET","depth":10,"bounds":{"left":0.79305184,"top":0.76576215,"width":0.08194814,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Thursdays at 10am ET / 4pm CET","depth":11,"bounds":{"left":0.79305184,"top":0.76576215,"width":0.08194814,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-4344484337332719459
|
-2831913820834514030
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Log In.
Log In.
[EMAIL]
Password
Login
Google
Forgot password?
Reset
Login via SAML SSO
Salaries, goals, expectations, technology…
We’d love to hear from you!
Take the survey
Take the survey
Did you know?
Did you know?
Planhat runs live training sessions
every
Tuesday
and
Thursday!
SIGN UP AT THE LINKS BELOW:
SIGN UP AT THE LINKS BELOW:
Tuesdays at 11am ET / 5pm CET
Tuesdays at 11am ET / 5pm CET
Thursdays at 10am ET / 4pm CET
Thursdays at 10am ET / 4pm CET...
|
22684
|
NULL
|
NULL
|
NULL
|
|
22685
|
974
|
45
|
2026-05-12T07:17:43.408143+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570263408_m1.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-2219013471096925094
|
-4245720915524734258
|
click
|
hybrid
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:43+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
22683
|
NULL
|
NULL
|
NULL
|
|
22683
|
974
|
44
|
2026-05-12T07:17:42.665774+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570262665_m1.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:42+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
NULL
|
-7433642549702625708
|
NULL
|
click
|
ocr
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:42+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22684
|
975
|
58
|
2026-05-12T07:17:42.665760+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570262665_m2.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
TrassworeFavouritesE jiminny© Recents* Application TrassworeFavouritesE jiminny© Recents* Applications© DocumentsIcloud• iCloud Drive992 Svnc toldeLocationsDXP4800PLUS-B5FA® Network• CRM• Orange• Red• Yellow• Greer• Purple• All lags..VIewACCOUnTSWindowHelpv 2026= Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4i Daily 2026-05-08.mp4* 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4*s 1-1 2026-04-24.mp4Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4w Daily 2026-04-23.mp4n Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dau Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Daily 2026-04-17.mp4Fu Daily 2026-04-16.mp4ax Planning 2026-04-15.mp4. Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4= Daily 2026-04-09.mp4• Daily 2026-04-07.mp4- Daily 2026-04-06.mp4= Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4н: Retro 2026-03-31.mp4Daily 2026-03-31.mp4•: Refinement 2026-03-30.m04Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4- Refinenment 2026-03-23.mp4- Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4Daily 2026-03-20.mp4an Planing 2026-03-18-converted.mp4= Retinement 2026-02-09-converted.mo4)eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/1-1 2026-03-12.mp4Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/- Daily 2026-03-10.mp4w Refinement 2026-03-09.mp4na Daily 2026-03-09.mo4Po Dailv 2026-03-06.m04<40 ll Support Daily - in 4h 43 m100% C42 & • Tue 12 May 10:17:42• -app-eu3.planhat.com/logiiNew Tab~ Google GeminiJy 20820 es reindex stream mode[JY-20725] [HubSpot) Optimise Cie Search in Jiminnv2results for "olanh"@a [EMAIL] log in (planhat.i...JiminnyB ProfileAll Items? FavoritesÄ Watchtower< Developen@ Employee@ EngineeringIntearation Accounts7 JiminnyTAGS• 2FA• CSV Import 28.01.22• LastPass Import 1.11.21oantervi2 Jiminnv@ Employee [EMAIL]:/app-eus.olanhat.com•Last edited Wednescav. Juv 5. 2023 at 3:29:20 PMIplanhatHelo+ New ItemU Share0 EditCopy.og [EMAIL] password? Resetn via SAML SSOHow has CustomerSuccess changed inthe past five years?Salaries, goals, expectations, technology….We'd love to hear from you!Take the surveyDid you know?Planhat runs live training sessionsevery Tuesday and Thursday!SIGN LID AT THE LINKS REIOW.Tuesdays at 11am ET / 5pm CETThursdays at 10am ET / 4pm CETtimestamoed soeaker reterencesO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini isAlcan mike micnkec co dolidle chockt Yolr orivaev domiinSummarize page...
|
NULL
|
-5876430447048297888
|
NULL
|
click
|
ocr
|
NULL
|
TrassworeFavouritesE jiminny© Recents* Application TrassworeFavouritesE jiminny© Recents* Applications© DocumentsIcloud• iCloud Drive992 Svnc toldeLocationsDXP4800PLUS-B5FA® Network• CRM• Orange• Red• Yellow• Greer• Purple• All lags..VIewACCOUnTSWindowHelpv 2026= Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4i Daily 2026-05-08.mp4* 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4*s 1-1 2026-04-24.mp4Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4w Daily 2026-04-23.mp4n Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dau Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Daily 2026-04-17.mp4Fu Daily 2026-04-16.mp4ax Planning 2026-04-15.mp4. Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4= Daily 2026-04-09.mp4• Daily 2026-04-07.mp4- Daily 2026-04-06.mp4= Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4н: Retro 2026-03-31.mp4Daily 2026-03-31.mp4•: Refinement 2026-03-30.m04Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4- Refinenment 2026-03-23.mp4- Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4Daily 2026-03-20.mp4an Planing 2026-03-18-converted.mp4= Retinement 2026-02-09-converted.mo4)eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/1-1 2026-03-12.mp4Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/- Daily 2026-03-10.mp4w Refinement 2026-03-09.mp4na Daily 2026-03-09.mo4Po Dailv 2026-03-06.m04<40 ll Support Daily - in 4h 43 m100% C42 & • Tue 12 May 10:17:42• -app-eu3.planhat.com/logiiNew Tab~ Google GeminiJy 20820 es reindex stream mode[JY-20725] [HubSpot) Optimise Cie Search in Jiminnv2results for "olanh"@a [EMAIL] log in (planhat.i...JiminnyB ProfileAll Items? FavoritesÄ Watchtower< Developen@ Employee@ EngineeringIntearation Accounts7 JiminnyTAGS• 2FA• CSV Import 28.01.22• LastPass Import 1.11.21oantervi2 Jiminnv@ Employee [EMAIL]:/app-eus.olanhat.com•Last edited Wednescav. Juv 5. 2023 at 3:29:20 PMIplanhatHelo+ New ItemU Share0 EditCopy.og [EMAIL] password? Resetn via SAML SSOHow has CustomerSuccess changed inthe past five years?Salaries, goals, expectations, technology….We'd love to hear from you!Take the surveyDid you know?Planhat runs live training sessionsevery Tuesday and Thursday!SIGN LID AT THE LINKS REIOW.Tuesdays at 11am ET / 5pm CETThursdays at 10am ET / 4pm CETtimestamoed soeaker reterencesO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini isAlcan mike micnkec co dolidle chockt Yolr orivaev domiinSummarize page...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22682
|
975
|
57
|
2026-05-12T07:17:41.128136+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570261128_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.575419,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.5865922,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5826017,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.6097366,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
3685474980799444353
|
-2831911623152649326
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen...
|
22680
|
NULL
|
NULL
|
NULL
|
|
22681
|
974
|
43
|
2026-05-12T07:17:41.107808+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570261107_m1.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-6948516146508445226
|
-3020487686398985585
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search...
|
22677
|
NULL
|
NULL
|
NULL
|
|
22680
|
975
|
56
|
2026-05-12T07:17:40.502188+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570260502_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Log In.
Log In.
[EMAIL]
Password
Login
Google
Forgot password?
Reset
Login via SAML SSO
Salaries, goals, expectations, technology…
We’d love to hear from you!
Take the survey
Take the survey
Did you know?
Did you know?
Planhat runs live training sessions
every
Tuesday
and
Thursday!
SIGN UP AT THE LINKS BELOW:
SIGN UP AT THE LINKS BELOW:
Tuesdays at 11am ET / 5pm CET
Tuesdays at 11am ET / 5pm CET
Thursdays at 10am ET / 4pm CET
Thursdays at 10am ET / 4pm CET...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.575419,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.5865922,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5826017,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.6097366,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Log In.","depth":9,"bounds":{"left":0.51662236,"top":0.3180367,"width":0.15292554,"height":0.06703911},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Log In.","depth":10,"bounds":{"left":0.51662236,"top":0.3140463,"width":0.088597074,"height":0.075418994},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"lukas.kovalik@jiminny.com","depth":9,"bounds":{"left":0.51662236,"top":0.40822026,"width":0.13297872,"height":0.04868316},"on_screen":true,"value":"lukas.kovalik@jiminny.com","help_text":"","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXTextField","text":"Password","depth":9,"bounds":{"left":0.51662236,"top":0.47685555,"width":0.13297872,"height":0.04868316},"on_screen":true,"help_text":"","role_description":"secure text field","subrole":"AXSecureTextField","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Login","depth":10,"bounds":{"left":0.51662236,"top":0.5734238,"width":0.059840426,"height":0.04868316},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXLink","text":"Google","depth":9,"bounds":{"left":0.5831117,"top":0.5734238,"width":0.020279255,"height":0.04868316},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Forgot password?","depth":10,"bounds":{"left":0.51662236,"top":0.6472466,"width":0.044215426,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Reset","depth":11,"bounds":{"left":0.56083775,"top":0.6472466,"width":0.013630319,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Login via SAML SSO","depth":11,"bounds":{"left":0.51662236,"top":0.6775738,"width":0.04870346,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Salaries, goals, expectations, technology…","depth":11,"bounds":{"left":0.77726066,"top":0.44852355,"width":0.11353058,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"We’d love to hear from you!","depth":11,"bounds":{"left":0.7972075,"top":0.4708699,"width":0.07363697,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Take the survey","depth":10,"bounds":{"left":0.7642952,"top":0.5159617,"width":0.13962767,"height":0.04868316},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Take the survey","depth":11,"bounds":{"left":0.8121675,"top":0.5323224,"width":0.043716755,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Did you know?","depth":9,"bounds":{"left":0.76163566,"top":0.6300878,"width":0.14494681,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Did you know?","depth":10,"bounds":{"left":0.80701464,"top":0.6312849,"width":0.054022606,"height":0.023144454},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Planhat runs live training sessions","depth":10,"bounds":{"left":0.7875665,"top":0.66001594,"width":0.09291888,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"every","depth":10,"bounds":{"left":0.79238695,"top":0.6823623,"width":0.01662234,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Tuesday","depth":10,"bounds":{"left":0.8090093,"top":0.6823623,"width":0.024268618,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and","depth":10,"bounds":{"left":0.83327794,"top":0.6823623,"width":0.013464096,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thursday!","depth":10,"bounds":{"left":0.84674203,"top":0.6823623,"width":0.028922873,"height":0.01715882},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"SIGN UP AT THE LINKS BELOW:","depth":9,"bounds":{"left":0.76163566,"top":0.73064643,"width":0.14494681,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SIGN UP AT THE LINKS BELOW:","depth":10,"bounds":{"left":0.80285907,"top":0.7302474,"width":0.062333778,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Tuesdays at 11am ET / 5pm CET","depth":10,"bounds":{"left":0.7947141,"top":0.74660814,"width":0.07862367,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Tuesdays at 11am ET / 5pm CET","depth":11,"bounds":{"left":0.7947141,"top":0.74660814,"width":0.07862367,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Thursdays at 10am ET / 4pm CET","depth":10,"bounds":{"left":0.79305184,"top":0.76576215,"width":0.08194814,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Thursdays at 10am ET / 4pm CET","depth":11,"bounds":{"left":0.79305184,"top":0.76576215,"width":0.08194814,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-4344484337332719459
|
-2831913820834514030
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Log In.
Log In.
[EMAIL]
Password
Login
Google
Forgot password?
Reset
Login via SAML SSO
Salaries, goals, expectations, technology…
We’d love to hear from you!
Take the survey
Take the survey
Did you know?
Did you know?
Planhat runs live training sessions
every
Tuesday
and
Thursday!
SIGN UP AT THE LINKS BELOW:
SIGN UP AT THE LINKS BELOW:
Tuesdays at 11am ET / 5pm CET
Tuesdays at 11am ET / 5pm CET
Thursdays at 10am ET / 4pm CET
Thursdays at 10am ET / 4pm CET...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22679
|
974
|
42
|
2026-05-12T07:17:39.157778+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570259157_m1.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:38+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
NULL
|
-3010415061922523786
|
NULL
|
click
|
ocr
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:38+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
22677
|
NULL
|
NULL
|
NULL
|
|
22677
|
974
|
41
|
2026-05-12T07:17:38.549589+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570258549_m1.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:38+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
NULL
|
-3010415061922523786
|
NULL
|
click
|
ocr
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:38+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22678
|
975
|
55
|
2026-05-12T07:17:38.541877+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570258541_m2.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
TrassworeFavouritesE jiminny© Recents* Application TrassworeFavouritesE jiminny© Recents* Applications© DocumentsIcloud• iCloud Drive992 Svnc toldeLocationsDXP4800PLUS-B5FA® Network• CRM• Orange• Red• Yellow• Greer• Purple• All lags..VIewACCOUnTSWindowHelpv 2026= Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4a Daily 2026-05-08.mp4* 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4*s 1-1 2026-04-24.mp4Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4w Daily 2026-04-23.mp4n Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dau Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Daily 2026-04-17.mp4Fu Daily 2026-04-16.mp4ax Planning 2026-04-15.mp4. Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4= Daily 2026-04-09.mp4a Daily 2026-04-07.mp4- Daily 2026-04-06.mp4= Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4н: Retro 2026-03-31.mp4Daily 2026-03-31.mp4•: Refinement 2026-03-30.mo4Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4- Refinenment 2026-03-23.mp4- Daily 2026-03-23.mp4• BE chaoter 2026-03-20.mo4Daily 2026-03-20.mp4an Planing 2026-03-18-converted.mp4= Retinement 2026-02-09-converted.mo4)eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/1-1 2026-03-12.mp4Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/- Daily 2026-03-10.mp4w Refinement 2026-03-09.mp4na Daily 2026-03-09.mo4Po Dailv 2026-03-06.m04<40 ll Support Daily - in 4h 43 m100% C4 & • Tue 12 May 10:17:38JiminnyB ProfileAll Items? FavoritesÄ Watchtower< Developen@ Employee@ EngineeringIntearation Accounts7 JiminnyTAGS• 2FA• CSV Import 28.01.22• LastPass Import 1.11.21oantervi• -app-eu3.planhat.com/loginNew Tab~ Google GeminiJy 20820 es reindex stream mode[JY-20725] [HubSpot) Optimise Cie Search in Jiminnv2results for "olanh"@a 152 Jiminnv@ Employee [EMAIL] log in (planhat.i...lukas.kovallkaminny.comhttos:/app-eus.olanhat.complanhatHelo+ New ItemU Share0 EditCopyGoodo•Last edited Wednescav. Juv 5. 2023 at 3:29:20 PMI.og In.mailasswordLocinot password? Resetn via SAML SSOHow has CustomerSuccess changed inthe past five years?Salaries, goals, expectations, technology….We'd love to hear from you!Take the surveyDid you know?Planhat runs live training sessionsevery Tuesday and Thursday!SIGN LID AT THE LINKS REIOW.Tuesdays at 11am ET / 5pm CETThursdays at 10am ET / 4pm CETtimestamped speaker referencesO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini isJlrcan make micnkes coldollidle chockt Yottr drivaeu wdomtnSummarize page...
|
NULL
|
5936077684047871760
|
NULL
|
click
|
ocr
|
NULL
|
TrassworeFavouritesE jiminny© Recents* Application TrassworeFavouritesE jiminny© Recents* Applications© DocumentsIcloud• iCloud Drive992 Svnc toldeLocationsDXP4800PLUS-B5FA® Network• CRM• Orange• Red• Yellow• Greer• Purple• All lags..VIewACCOUnTSWindowHelpv 2026= Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4a Daily 2026-05-08.mp4* 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4*s 1-1 2026-04-24.mp4Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4w Daily 2026-04-23.mp4n Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dau Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Daily 2026-04-17.mp4Fu Daily 2026-04-16.mp4ax Planning 2026-04-15.mp4. Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4= Daily 2026-04-09.mp4a Daily 2026-04-07.mp4- Daily 2026-04-06.mp4= Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4н: Retro 2026-03-31.mp4Daily 2026-03-31.mp4•: Refinement 2026-03-30.mo4Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4- Refinenment 2026-03-23.mp4- Daily 2026-03-23.mp4• BE chaoter 2026-03-20.mo4Daily 2026-03-20.mp4an Planing 2026-03-18-converted.mp4= Retinement 2026-02-09-converted.mo4)eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/1-1 2026-03-12.mp4Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/- Daily 2026-03-10.mp4w Refinement 2026-03-09.mp4na Daily 2026-03-09.mo4Po Dailv 2026-03-06.m04<40 ll Support Daily - in 4h 43 m100% C4 & • Tue 12 May 10:17:38JiminnyB ProfileAll Items? FavoritesÄ Watchtower< Developen@ Employee@ EngineeringIntearation Accounts7 JiminnyTAGS• 2FA• CSV Import 28.01.22• LastPass Import 1.11.21oantervi• -app-eu3.planhat.com/loginNew Tab~ Google GeminiJy 20820 es reindex stream mode[JY-20725] [HubSpot) Optimise Cie Search in Jiminnv2results for "olanh"@a 152 Jiminnv@ Employee [EMAIL] log in (planhat.i...lukas.kovallkaminny.comhttos:/app-eus.olanhat.complanhatHelo+ New ItemU Share0 EditCopyGoodo•Last edited Wednescav. Juv 5. 2023 at 3:29:20 PMI.og In.mailasswordLocinot password? Resetn via SAML SSOHow has CustomerSuccess changed inthe past five years?Salaries, goals, expectations, technology….We'd love to hear from you!Take the surveyDid you know?Planhat runs live training sessionsevery Tuesday and Thursday!SIGN LID AT THE LINKS REIOW.Tuesdays at 11am ET / 5pm CETThursdays at 10am ET / 4pm CETtimestamped speaker referencesO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini isJlrcan make micnkes coldollidle chockt Yottr drivaeu wdomtnSummarize page...
|
22676
|
NULL
|
NULL
|
NULL
|
|
22675
|
974
|
40
|
2026-05-12T07:17:34.934538+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570254934_m1.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-1991970465242606427
|
-2732256210735629693
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu...
|
22673
|
NULL
|
NULL
|
NULL
|
|
22676
|
975
|
54
|
2026-05-12T07:17:34.934513+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570254934_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.575419,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.5865922,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5826017,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.6097366,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
3921098929698955868
|
-2831913821102687594
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22674
|
975
|
53
|
2026-05-12T07:17:33.242908+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570253242_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
about:newtab
New Tab
New Tab
Jy 20820 es r New Tab
about:newtab
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
Close tab
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score...
|
[{"role":"AXStaticText","text& [{"role":"AXStaticText","text":"New Tab","depth":4,"bounds":{"left":0.32696143,"top":0.5826017,"width":0.015458777,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"about:newtab","depth":4,"bounds":{"left":0.32696143,"top":0.5933759,"width":0.023936171,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.575419,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.5865922,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5826017,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.60814047,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.61931366,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.61532325,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.6097366,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
2680716571698625297
|
-2831911623152649326
|
click
|
accessibility
|
NULL
|
New Tab
about:newtab
New Tab
New Tab
Jy 20820 es r New Tab
about:newtab
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
Close tab
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score...
|
22672
|
NULL
|
NULL
|
NULL
|
|
22673
|
974
|
39
|
2026-05-12T07:17:33.233293+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570253233_m1.jpg...
|
Firefox
|
Login — Work
|
True
|
app-eu3.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
about:newtab
New Tab
New Tab
Jy 20820 es r New Tab
about:newtab
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
Close tab
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close...
|
[{"role":"AXStaticText","text& [{"role":"AXStaticText","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"about:newtab","depth":4,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
646656717547085956
|
-2444025834583934326
|
click
|
accessibility
|
NULL
|
New Tab
about:newtab
New Tab
New Tab
Jy 20820 es r New Tab
about:newtab
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
Close tab
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22672
|
975
|
52
|
2026-05-12T07:17:31.365614+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570251365_m2.jpg...
|
Firefox
|
Work — Mozilla Firefox
|
True
|
app-eu3.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3
app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Read app-eu3.planhat.com...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.575419,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.5865922,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3","depth":4,"bounds":{"left":0.24335106,"top":0.60814047,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3","depth":5,"bounds":{"left":0.25664893,"top":0.61931366,"width":0.25681517,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.61532325,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.6424581,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Read app-eu3.planhat.com","depth":5,"bounds":{"left":0.46010637,"top":0.9876297,"width":0.047041222,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-1740739471292033851
|
-2831913822176429422
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3
app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Read app-eu3.planhat.com...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22671
|
974
|
38
|
2026-05-12T07:17:28.775755+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570248775_m1.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:28+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
NULL
|
-1670641185697820209
|
NULL
|
click
|
ocr
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:28+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
22668
|
NULL
|
NULL
|
NULL
|
|
22670
|
975
|
51
|
2026-05-12T07:17:28.316203+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570248316_m2.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
TrassworeFavouritesjiminny• Recents* Applications| TrassworeFavouritesjiminny• Recents* Applications|9 DocumentsIcloud• iCloud Drive992 Svnc toldeLocations0 DXP4800PLUS-B5F A49 Network• CRM• Orange• Red• Yellow• Greer• Purple• All lags..VIewACCOUnTSWindowMelpv 2026• Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4a Daily 2026-05-08.mp4Iя 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4ws 1-1 2026-04-24.mp4= Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4nDaily 2026-04-22 mnлwn Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4D Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Daily 2026-04-17.mp4ru Daily 2026-04-16.mp4wa Planning 2026-04-15.mp4. Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4• Daily 2026-04-09.mp4Daily 2026-04-07.mp4Daily 2026-04-06.mp4= Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4l: Retro 2026-03-31.mp4Daily 2026-03-31.mp4•: Refinement 2026-03-30.mo4m Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4• Refineiment 2026-03-23.mp4-• Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4• Daily 2026-03-20.mp4am Planing 2026-03-18-converted.mp4#. Refinement 2026-02-09-converted.mp4eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/ra 1-1 2026-03-12.mpFa Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/Dailv 2026-03-10.mp4xI Refinement 2026-03-09.mp4na Daily 2026-03-09.mo4н лail aлe лe лe meл••0 ШImport bookmarks... Sprint Board• SRD Queue- New Tab• Google GeminiJy 20820 es reindex stream modee Search in Jiminnv2results for "olanh"PlannatSalesforce log in (planhat.i..O Circle CI & PRODUS J StagingHeloNew ItemU Shareo EditPull requests . limin…Workers | Datadog M Jiminny Mail Dashboards | Datad..Jiminny¿ Profile-All Items? FavoritesWatchtower< Developeno. =moloveeEngineeringIntearation Accounts7 JiminnyTAGSO 2FA• CSV Import 28.01.22(• LastPacs Imnort 111.21|oantervi2 Jiminnv(o) Emolovee [EMAIL]:/app-eus.olanhat.com•Last edited Wednescav. Juv 5. 2023 at 3:29:20 PMIGoodFirefoxG Search with Google or enter addressPlatform Sarint302-..liminnviminnylappInboy (1 620) -lukas.kovalik…liminnvbrief justification, and up to threeO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini isAlcan mike micnkec co dolidle chockt Yolr orivaev domiinSummarize page"suppont Dally • In 4n 43 m100% 2• Tue 12 May 10:17:271Y-20542 addlAJ reports.Mept - DailvPlattorm...
|
NULL
|
-554091232307711275
|
NULL
|
visual_change
|
ocr
|
NULL
|
TrassworeFavouritesjiminny• Recents* Applications| TrassworeFavouritesjiminny• Recents* Applications|9 DocumentsIcloud• iCloud Drive992 Svnc toldeLocations0 DXP4800PLUS-B5F A49 Network• CRM• Orange• Red• Yellow• Greer• Purple• All lags..VIewACCOUnTSWindowMelpv 2026• Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4a Daily 2026-05-08.mp4Iя 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4ws 1-1 2026-04-24.mp4= Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4nDaily 2026-04-22 mnлwn Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4D Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Daily 2026-04-17.mp4ru Daily 2026-04-16.mp4wa Planning 2026-04-15.mp4. Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4• Daily 2026-04-09.mp4Daily 2026-04-07.mp4Daily 2026-04-06.mp4= Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4l: Retro 2026-03-31.mp4Daily 2026-03-31.mp4•: Refinement 2026-03-30.mo4m Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4• Refineiment 2026-03-23.mp4-• Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4• Daily 2026-03-20.mp4am Planing 2026-03-18-converted.mp4#. Refinement 2026-02-09-converted.mp4eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/ra 1-1 2026-03-12.mpFa Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/Dailv 2026-03-10.mp4xI Refinement 2026-03-09.mp4na Daily 2026-03-09.mo4н лail aлe лe лe meл••0 ШImport bookmarks... Sprint Board• SRD Queue- New Tab• Google GeminiJy 20820 es reindex stream modee Search in Jiminnv2results for "olanh"PlannatSalesforce log in (planhat.i..O Circle CI & PRODUS J StagingHeloNew ItemU Shareo EditPull requests . limin…Workers | Datadog M Jiminny Mail Dashboards | Datad..Jiminny¿ Profile-All Items? FavoritesWatchtower< Developeno. =moloveeEngineeringIntearation Accounts7 JiminnyTAGSO 2FA• CSV Import 28.01.22(• LastPacs Imnort 111.21|oantervi2 Jiminnv(o) Emolovee [EMAIL]:/app-eus.olanhat.com•Last edited Wednescav. Juv 5. 2023 at 3:29:20 PMIGoodFirefoxG Search with Google or enter addressPlatform Sarint302-..liminnviminnylappInboy (1 620) -lukas.kovalik…liminnvbrief justification, and up to threeO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini isAlcan mike micnkec co dolidle chockt Yolr orivaev domiinSummarize page"suppont Dally • In 4n 43 m100% 2• Tue 12 May 10:17:271Y-20542 addlAJ reports.Mept - DailvPlattorm...
|
22669
|
NULL
|
NULL
|
NULL
|
|
22669
|
975
|
50
|
2026-05-12T07:17:27.148657+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570247148_m2.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
TrassworoFavouritesjiminny• Recents* Applications9 TrassworoFavouritesjiminny• Recents* Applications9 Documentsii lukasIcloud• iCloud Drive992 Svnc toldeLocations0 DXP4800PLUS-B5F A49 NetworkTags• CRM• Orange• Red• Yellow• Purple|• All lagsVIewAccountsWindowMelpv 2026• Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4a Daily 2026-05-08.nIя 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4ws 1-1 2026-04-24.mp4= Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4nDaily 2026-04-22 mn/wn Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dr Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4L Daily 2026-04-17.mo4ru Daily 2026-04-16.mp4Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4• Daily 2026-04-09.mp4Daily 2026-04-07.mp4Daily 2026-04-06.mp4=Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4l: Retro 2026-03-31.mp4Daily 2026-03-31.mp4: Refinement 2026-03-30.mo4m Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4• Refineiment 2026-03-23.mp4-• Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4• Daily 2026-03-20.mp4am Planing 2026-03-18-converted.mp4_Retinement 2026-02-09-converted.mo4eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/ra 1-1 2026-03-12.mpFa Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/Dailv 2026-03-10.mp4xI Refinement 2026-03-09.mp4na Daily 2026-03-09.mo4tanailu bane na.ne mnllJiminny¿ Profile-All Items? FavoritesWatchtower< DevelopenVAULTSo. =moloveeEngineeringIntearation Accounts7 JiminnvTAGSO 2FA• CSV Import 28.01.22(• LastPacs Imnort 111.21.oantervi••0 ШImport bookmarks... Sprint Board• SRD QueueO Circle CI & PROD USJ Staging- New Tab• Google GeminiJy 20820 es reindex stream modeeolanhHeloNew Item= Planhat — [EMAIL] app-eu3.planhat.com20 results 1) Share0 Edit• Salestorce log in ([EMAIL]) - integration-accoun.• planh — Show all matching itemsasier and more secure thanMS Dynamics Pro [EMAIL] [EMAIL] transcription pa….FantasticMicrosoft transcription [URL_WITH_CREDENTIALS] justification, and up to three• Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini tAlcan mike micnkec co dolidle chockt Yolr orivaev domiinSummarize pageWorkers | Datadog M Jiminny Mail & Dashboards | Datad..Firefoxsearch with Gooqle or enter addressPlatform Sarint302-..liminntiminnylappInboy (1 620) -lukas.kovalik…liminnv"suppont Dally • In 4n 43m100% Lz• Tue 12 May 10:17:261Y-20542 addlAJ reports.Mept - DailvPlattorm...
|
NULL
|
-5354076051370249108
|
NULL
|
click
|
ocr
|
NULL
|
TrassworoFavouritesjiminny• Recents* Applications9 TrassworoFavouritesjiminny• Recents* Applications9 Documentsii lukasIcloud• iCloud Drive992 Svnc toldeLocations0 DXP4800PLUS-B5F A49 NetworkTags• CRM• Orange• Red• Yellow• Purple|• All lagsVIewAccountsWindowMelpv 2026• Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4a Daily 2026-05-08.nIя 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4ws 1-1 2026-04-24.mp4= Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4nDaily 2026-04-22 mn/wn Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dr Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4L Daily 2026-04-17.mo4ru Daily 2026-04-16.mp4Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4• Daily 2026-04-09.mp4Daily 2026-04-07.mp4Daily 2026-04-06.mp4=Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4l: Retro 2026-03-31.mp4Daily 2026-03-31.mp4: Refinement 2026-03-30.mo4m Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4• Refineiment 2026-03-23.mp4-• Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4• Daily 2026-03-20.mp4am Planing 2026-03-18-converted.mp4_Retinement 2026-02-09-converted.mo4eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/ra 1-1 2026-03-12.mpFa Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/Dailv 2026-03-10.mp4xI Refinement 2026-03-09.mp4na Daily 2026-03-09.mo4tanailu bane na.ne mnllJiminny¿ Profile-All Items? FavoritesWatchtower< DevelopenVAULTSo. =moloveeEngineeringIntearation Accounts7 JiminnvTAGSO 2FA• CSV Import 28.01.22(• LastPacs Imnort 111.21.oantervi••0 ШImport bookmarks... Sprint Board• SRD QueueO Circle CI & PROD USJ Staging- New Tab• Google GeminiJy 20820 es reindex stream modeeolanhHeloNew Item= Planhat — [EMAIL] app-eu3.planhat.com20 results 1) Share0 Edit• Salestorce log in ([EMAIL]) - integration-accoun.• planh — Show all matching itemsasier and more secure thanMS Dynamics Pro [EMAIL] [EMAIL] transcription pa….FantasticMicrosoft transcription [URL_WITH_CREDENTIALS] justification, and up to three• Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini tAlcan mike micnkec co dolidle chockt Yolr orivaev domiinSummarize pageWorkers | Datadog M Jiminny Mail & Dashboards | Datad..Firefoxsearch with Gooqle or enter addressPlatform Sarint302-..liminntiminnylappInboy (1 620) -lukas.kovalik…liminnv"suppont Dally • In 4n 43m100% Lz• Tue 12 May 10:17:261Y-20542 addlAJ reports.Mept - DailvPlattorm...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22668
|
974
|
37
|
2026-05-12T07:17:27.136572+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570247136_m1.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:26+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
NULL
|
-8567104880481717845
|
NULL
|
click
|
ocr
|
NULL
|
QuickTime PlayerFileEditViewWindowHelpPetko Kashin QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 43 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:17:26+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22667
|
975
|
49
|
2026-05-12T07:17:25.315399+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570245315_m2.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
TrassworoFavouritesjiminny• Recents* Applications9 TrassworoFavouritesjiminny• Recents* Applications9 DocumentsIcloud• iCloud Drive992 Svnc toldelLocations0 DXP4800PLUS-B5FA49 NetworkTags• CRM• Orange• Red• Yellow• Purple|• All lags.VIewAccountsWindowhelv 2026• Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4wu Daily 2026-05-08.mp4Iя 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4us 1-1 2026-04-24.mp4= Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4nDaily 2026-04-22 mnлwn Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dr Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Dailv 2026-04-17.mp4ru Daily 2026-04-16.mp4Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4• Daily 2026-04-09.mp4Daily 2026-04-07.mp4Daily 2026-04-06.mp4=Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4l: Retro 2026-03-31.mp4m Daily 2026-03-31.mp4: Refinement 2026-03-30.mo4m Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4• Refineiment 2026-03-23.mp4-• Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4• Daily 2026-03-20.mp4am Planing 2026-03-18-converted.mp4= Retinement 2026-02-09-converted.mo4)eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4• Planina 2026-03-18.mo4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4= Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/ra 1-1 2026-03-12.mp/Fa Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/Dailv 2026-03-10.mp4xI Refinement 2026-03-09.mp4e Dailv 2026-03-09 mo4ta nailu bane na.ne mnлlJiminny¿ Profile-All Items? FavoritesWatchtower< DevelopenVAULTSo. =moloveeEngineeringIntearation Accounts7 JiminnyTAGSO 2FA• CSV Import 28.01.22(• LastPacs Imnort 111.21|oantervit••0 ШImport bookmarks... Sprint Board• SRD QueueO Circle CI & PROD USJ Staging- New Tab• Google GeminiJy 20820 es reindex stream modeeolanlheloNew Item= Planhat — [EMAIL] app-eu3.planhat.com20 results 1Shareo E[URL_WITH_CREDENTIALS] justification, and up to threeO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini tAlrcan mike micnkec co dolidle chocktYolr orivaev domiinSummarize page<S0 Ma"suppont Dally • In 4h 43 m100% Lz• Tue 12 May 10:17:24Workers | Datadog M Jiminny Mail & Dashboards | Datad..FirefoxG Search with Google or enter addressPlatform Sarint302 -..liminntiminnylappInboy (1 620) -lukas.kovalik…liminnv1Y-20542 addlAJ reports.Mept - DailvPlattorm...
|
NULL
|
-2684762608853232682
|
NULL
|
visual_change
|
ocr
|
NULL
|
TrassworoFavouritesjiminny• Recents* Applications9 TrassworoFavouritesjiminny• Recents* Applications9 DocumentsIcloud• iCloud Drive992 Svnc toldelLocations0 DXP4800PLUS-B5FA49 NetworkTags• CRM• Orange• Red• Yellow• Purple|• All lags.VIewAccountsWindowhelv 2026• Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4wu Daily 2026-05-08.mp4Iя 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4us 1-1 2026-04-24.mp4= Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4nDaily 2026-04-22 mnлwn Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dr Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Dailv 2026-04-17.mp4ru Daily 2026-04-16.mp4Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4• Daily 2026-04-09.mp4Daily 2026-04-07.mp4Daily 2026-04-06.mp4=Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4l: Retro 2026-03-31.mp4m Daily 2026-03-31.mp4: Refinement 2026-03-30.mo4m Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4• Refineiment 2026-03-23.mp4-• Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4• Daily 2026-03-20.mp4am Planing 2026-03-18-converted.mp4= Retinement 2026-02-09-converted.mo4)eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4• Planina 2026-03-18.mo4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4= Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/ra 1-1 2026-03-12.mp/Fa Daily 2026-03-12.mp4aa Daily 2026-03-11.mn/Dailv 2026-03-10.mp4xI Refinement 2026-03-09.mp4e Dailv 2026-03-09 mo4ta nailu bane na.ne mnлlJiminny¿ Profile-All Items? FavoritesWatchtower< DevelopenVAULTSo. =moloveeEngineeringIntearation Accounts7 JiminnyTAGSO 2FA• CSV Import 28.01.22(• LastPacs Imnort 111.21|oantervit••0 ШImport bookmarks... Sprint Board• SRD QueueO Circle CI & PROD USJ Staging- New Tab• Google GeminiJy 20820 es reindex stream modeeolanlheloNew Item= Planhat — [EMAIL] app-eu3.planhat.com20 results 1Shareo E[URL_WITH_CREDENTIALS] justification, and up to threeO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini tAlrcan mike micnkec co dolidle chocktYolr orivaev domiinSummarize page<S0 Ma"suppont Dally • In 4h 43 m100% Lz• Tue 12 May 10:17:24Workers | Datadog M Jiminny Mail & Dashboards | Datad..FirefoxG Search with Google or enter addressPlatform Sarint302 -..liminntiminnylappInboy (1 620) -lukas.kovalik…liminnv1Y-20542 addlAJ reports.Mept - DailvPlattorm...
|
22666
|
NULL
|
NULL
|
NULL
|
|
22666
|
975
|
48
|
2026-05-12T07:17:22.360448+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570242360_m2.jpg...
|
iTerm2
|
NULL
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
TrassworeFavouritesjiminny@ Recents* Applications| TrassworeFavouritesjiminny@ Recents* Applications|9 Document.Icloud• iCloud Drive992 Svnc toldeLocations0 DXP4800PLUS-B5F A49 Network• CRMI• Orange• Red• Yellow• Greer• Purple|• All lags..CS0 Ma"suppont Dally • In 4n 43 m100% Lz• Tue 12 May 10:17:21VIewAccountsWindowheltv 2026• Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4a Daily 2026-05-08.nIя 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4us 1-1 2026-04-24.mp4= Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4# Daily 2026-04-22 mn/wn Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dr Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Daily 2026-04-17.mp4ru Daily 2026-04-16.mp4wa Planning 2026-04-15.mp4. Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4• Daily 2026-04-09.mp4Dailv 2026-04-08.mo4Daily 2026-04-07.mp4=» Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4Dally 2020-03-31.mp4•: Refinement 2026-03-30.mo4m Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4• Refirment 2026-03-23.mp4-• Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4• Daily 2026-03-20.mp4am Planing 2026-03-18-converted.mp4#. Refinement 2026-02-09-converted.mp4eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/кы 1-1 2026-03-12.mр4Dailv 2026-03-12.mp4aa Daily 2026-03-11.mn/• Refinement 2026-03-09.mp4e Dailv 2026-03-09.mo4н лail aлe лe лe meлImport bookmarks... Sprint Board• SRD Queue- New Tab• Google GeminiJy 20820 es reindex stream modeO Circle CI & PRODUS J StagingPull requests. limin.Workers | Datadog M Jiminny Mail Dashboards | Datad...1Password1Password is trying to unlockTouch iD or enter [PASSWORD_FIELD] Account Pacsword.FirefoxEnter vour password1Password locked after 10 minutes of inactivityG Search with Google or enter addressPlatform Sarint302-..liminnviminnylappInboy (1 620) -lukas.kovalik…liminnv1Y-20542 addlAJ reports.Mept - DailvPlattormbrief justification, and up to threeO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini isAlcan mike micnkec co dolidle chockt Yolr orivaev domiinSummarize page...
|
NULL
|
4287400542967433804
|
NULL
|
visual_change
|
ocr
|
NULL
|
TrassworeFavouritesjiminny@ Recents* Applications| TrassworeFavouritesjiminny@ Recents* Applications|9 Document.Icloud• iCloud Drive992 Svnc toldeLocations0 DXP4800PLUS-B5F A49 Network• CRMI• Orange• Red• Yellow• Greer• Purple|• All lags..CS0 Ma"suppont Dally • In 4n 43 m100% Lz• Tue 12 May 10:17:21VIewAccountsWindowheltv 2026• Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4a Daily 2026-05-08.nIя 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4us 1-1 2026-04-24.mp4= Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4# Daily 2026-04-22 mn/wn Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dr Refinement 2026-04-20.mp4Dally 4020-04-20.Mp4Daily 2026-04-17.mp4ru Daily 2026-04-16.mp4wa Planning 2026-04-15.mp4. Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4• Daily 2026-04-09.mp4Dailv 2026-04-08.mo4Daily 2026-04-07.mp4=» Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4Dally 2020-03-31.mp4•: Refinement 2026-03-30.mo4m Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4• Refirment 2026-03-23.mp4-• Daily 2026-03-23.mp4• BE chanter 2026-03-20.mo4• Daily 2026-03-20.mp4am Planing 2026-03-18-converted.mp4#. Refinement 2026-02-09-converted.mp4eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/кы 1-1 2026-03-12.mр4Dailv 2026-03-12.mp4aa Daily 2026-03-11.mn/• Refinement 2026-03-09.mp4e Dailv 2026-03-09.mo4н лail aлe лe лe meлImport bookmarks... Sprint Board• SRD Queue- New Tab• Google GeminiJy 20820 es reindex stream modeO Circle CI & PRODUS J StagingPull requests. limin.Workers | Datadog M Jiminny Mail Dashboards | Datad...1Password1Password is trying to unlockTouch iD or enter [PASSWORD_FIELD] Account Pacsword.FirefoxEnter vour password1Password locked after 10 minutes of inactivityG Search with Google or enter addressPlatform Sarint302-..liminnviminnylappInboy (1 620) -lukas.kovalik…liminnv1Y-20542 addlAJ reports.Mept - DailvPlattormbrief justification, and up to threeO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini isAlcan mike micnkec co dolidle chockt Yolr orivaev domiinSummarize page...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22665
|
974
|
36
|
2026-05-12T07:17:22.141702+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570242141_m1.jpg...
|
coreautha
|
1Password
|
True
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
1Password
1Password is trying to unlock.
Touch ID 1Password
1Password is trying to unlock.
Touch ID or enter your password to allow this.
Use Account Password…
Cancel...
|
[{"role":"AXStaticText","text& [{"role":"AXStaticText","text":"1Password","depth":1,"on_screen":true,"role_description":"text"},{"role":"AXStaticText","text":"1Password is trying to unlock.","depth":1,"on_screen":true,"role_description":"text"},{"role":"AXStaticText","text":"Touch ID or enter your password to allow this.","depth":1,"on_screen":true,"role_description":"text"},{"role":"AXButton","text":"Use Account Password…","depth":1,"on_screen":true,"role_description":"button","is_enabled":true,"is_focused":true},{"role":"AXButton","text":"Cancel","depth":1,"on_screen":true,"role_description":"button","is_enabled":true,"is_focused":false}]...
|
2367845838743006480
|
6549041024830674611
|
visual_change
|
accessibility
|
NULL
|
1Password
1Password is trying to unlock.
Touch ID 1Password
1Password is trying to unlock.
Touch ID or enter your password to allow this.
Use Account Password…
Cancel...
|
22663
|
NULL
|
NULL
|
NULL
|
|
22664
|
975
|
47
|
2026-05-12T07:17:19.273624+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570239273_m2.jpg...
|
Alfred
|
Alfred
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
1pa
|
[{"role":"AXTextField","text&# [{"role":"AXTextField","text":"1pa","depth":1,"on_screen":true,"value":"1pa","help_text":"Alfred Search","role_description":"text field","is_enabled":true,"is_focused":true}]...
|
2874096003370199024
|
2874096003370199024
|
visual_change
|
hybrid
|
NULL
|
1pa
FirefoxcalVIewMistorbookmarksProtllesToolsWind 1pa
FirefoxcalVIewMistorbookmarksProtllesToolsWindowmelpFavouritesjiminny(®) AirDrop• RecentsA Applications|9 Documentsii lukasv 2026• Daily 2026-05-12.mp4_ PLanhat Petko interest event 2026-05-11.mp4• Dailv 2026-05-11.mo4wu Daily 2026-05-08.mp4Iя 1-1 2026-05-07.mp4& Dailv 2026-05-07mo4us 1-1 2026-04-24.mp4= Daily 2026-04-24.mp4User Pilot introduction Adi 2026-04-23.mp4nDaily 2026-04-22 mnлIcloud• iCloud Drive992 Svnc toldewn Refinement 2026-04-06.mp4• Dailv 2026-04-21.mo4Dr Refinement 2026-04-20.mp4Locations0 DXP4800PLUS-B5F A49 Network• CRM• Orange• Red• YellowDally 4020-04-20.Mp4L Daily 2026-04-17.mo4ru Daily 2026-04-16.mp4wa Planning 2026-04-15.mp4Retro 2026-04-14.mp4n Daily 2026-04-14.mn/• User pilot (Adi) 2026-04-09.mp4• Daily 2026-04-09.mp4Dailv 2026-04-08.mo4Daily 2026-04-07.mp4• Greer= Daily 2026-04-03.mp4lax Planning 2026-04-01 & task split.mp4• Purple• All lags— Dally 2020-03-31.m04•: Refinement 2026-03-30.mo4m Daily 2026-03-30.mp4= Dallv 2026-03-26.m04B Daily 2026-02-24 mn4• Refinement 2026-03-23.mp4-• Daily 2026-03-23.mp4• BE chaoter 2026-03-20.mo4• Daily 2026-03-20.mp4am Planing 2026-03-18-converted.mp4#. Refinement 2026-02-09-converted.mp4eR Daily 2026-02-10 mn/- Review 2026-03-18.mp4am Planing 2026-03-18.mp4F Retro 2026-03-17 mn4- Daily 2026-03-17.mp4Refinement 2026-03-16.mp4• Dailv 2026-03-16.mo4F Daily 2026.02-12 mn/ra 1-1 2026-03-12.mpDailv 2026-03-12.mp4aa Daily 2026-03-11.mn/Dailv 2026-03-10.mp4xI Refinement 2026-03-09.mp4na Daily 2026-03-09.mo4н лail aлe лe лe meлImport bookmarks... Sprint Board• SRD Queue- New Tab• Google GeminiJy 20820 es reindex stream modeJY-20725) HubSpot Optimise ©CJY-20725 add HS rate limit handli( Pipelines - jiminny/appPull requests • jiminny/app[JY-20773] User Pilot not receivin@ Jy-20773 fix user pilot tracking of4 [JY-20776] Automated report - seA TypeError: League\Flysystem\FilesTypeError: League\Flysystem|FilesPlatform Sprint 3 Q2 - Platform Te• JY-20625 | JY-20742 | MCP POC LG LLM pricing overview with the mo.Planhat• LoainNew Tab— New TabO Circle CI & PRODUS J StagingPull requests . limin.Workers | Datadog M Jiminny Mail Dashboards | Datad...mon oaoe"<tablitles.Y-20261: A66lca scores in ranorama oysteliyan-</tablitle> with"<selection>@@ -4,7 +4,7..Show thinking vHere is a summary of the changes in theprovided code diffAl Call Scorina Additions• Pre-computed Scores: Added strictinstructions to treat the ai call scoreas pre-computed data. Uhe system mustinterpret and summarize the providedscores rather than inventing olrecalculating them• Context Breakdown: * Short Context:Now displays the Al Scorecard name andthe overall Al Score (the average of per-rule scores).Full Context: Details theai call score obiect, breakinadown individual rules. Each ruleincludes a title, prompt, a 1-5 score, abrief justification, and up to threeO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini tAlcan mike micnkec co dolidle chockt Yolr orivaev domiinSummarize pageFirefoxG Search with Google or enter addressPlatform Sarint302-..liminntiminnylappInboy (1 620) -lukas.kovalik…liminnv"suppont Dally • In 4n 43 m100% Lz• Tue 12 May 10:17:181Y-20542 addlAJ reports.Mept - DailvPlattorm...
|
22662
|
NULL
|
NULL
|
NULL
|
|
22663
|
974
|
35
|
2026-05-12T07:17:19.154881+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570239154_m1.jpg...
|
Alfred
|
Alfred
|
True
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
1p
|
[{"role":"AXTextField","text&# [{"role":"AXTextField","text":"1p","depth":1,"bounds":{"left":0.26180556,"top":0.16777778,"width":0.4763889,"height":0.05888889},"on_screen":true,"value":"1p","help_text":"Alfred Search","role_description":"text field","is_enabled":true,"is_focused":true}]...
|
4995732740252791322
|
4995732740252791322
|
visual_change
|
hybrid
|
NULL
|
1p
QuickTime PlayerFileEditViewWindowHelpPetko Kas 1p
QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelpGreetiScorecandre;f wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb9АIKB• ChatPlayground Al...Jiminny - Calenda...M GI1+ Back83 App Center& UP > PH UX1Password.app/Applications/1Password.app0 EditorPa Runs8 DataJiminnE All apps& Created by meG Recentty most activeIntegrationsR AutomationsP Private appsAction this item to copy this number to the clipboard1Password Bookmark'Find and open a 1Password 1Click Bookmark(ahol• Support Daily - in 4h 43 mlQM Inbox=Nate R= AFBuildirUserp100% C78• Tue 12 May 10:17:18+ |%8• Mon 11 May 12:17New tr+Work. Petko•** XEnabled4 Webhook incomingA Webhook282883Event details8bcea7d0-160b- aTriggeranythingOmsTnopered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (&Leave...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22662
|
975
|
46
|
2026-05-12T07:17:16.242606+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570236242_m2.jpg...
|
Firefox
|
Work — Mozilla Firefox
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Login
ws.planhat.com
New Tab
New Tab
Jy 20820 es r Login
ws.planhat.com
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
New Tab
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:...
|
[{"role":"AXStaticText","text& [{"role":"AXStaticText","text":"Login","depth":4,"bounds":{"left":0.32696143,"top":0.57222664,"width":0.009973404,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ws.planhat.com","depth":4,"bounds":{"left":0.32696143,"top":0.5830008,"width":0.027260639,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.074221864,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.08539505,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.10694334,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.11811652,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.1396648,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.15083799,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.17238627,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.18355946,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.20510775,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.21628092,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.23782921,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.2490024,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.27055067,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.28172386,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.30327216,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.31444532,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.33599362,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.3471668,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.36871508,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.37988827,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.40143654,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.41260973,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.43415803,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.44533122,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4668795,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.47805268,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.49960095,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.51077414,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.5323224,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5434956,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.56504387,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.57621706,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.57222664,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.5977654,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.6089386,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.6049481,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.632083,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.07741421,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.07741421,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.12609737,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.12609737,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.12609737,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.12609737,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.16999201,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.17238627,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.20590582,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.21548285,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.21628092,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.21707901,"width":0.06648936,"height":0.78292096},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.3320032,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.21548285,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.34317636,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.34557062,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.38707104,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.38906625,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.39026338,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4517159,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.45331204,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.47964883,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.47964883,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.50159615,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.50039905,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.612929,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.612929,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.612929,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.612929,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.63367915,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.63367915,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.6544294,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.6544294,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.7047087,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.7047087,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.726656,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.7254589,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.858739,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.858739,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.98244214,"width":0.11502659,"height":0.01755786},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9840383,"width":0.08809841,"height":0.015961707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.046210106,"height":-0.010375142},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.09507979,"height":-0.010375142},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
1761320631709701341
|
-2831770885664293998
|
visual_change
|
accessibility
|
NULL
|
Login
ws.planhat.com
New Tab
New Tab
Jy 20820 es r Login
ws.planhat.com
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
New Tab
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22661
|
975
|
45
|
2026-05-12T07:17:15.258386+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570235258_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
1481586658186100718
|
-2876372498323113334
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat...
|
22659
|
NULL
|
NULL
|
NULL
|
|
22660
|
974
|
34
|
2026-05-12T07:17:15.258375+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570235258_m1.jpg...
|
Firefox
|
Login — Work
|
True
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false}]...
|
5508579009360439157
|
-2867365299068372342
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions....
|
22657
|
NULL
|
NULL
|
NULL
|
|
22659
|
975
|
44
|
2026-05-12T07:17:12.893860+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570232893_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
ws.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Welcome back","depth":9,"bounds":{"left":0.48603722,"top":0.30526736,"width":0.14494681,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Welcome back","depth":10,"bounds":{"left":0.48603722,"top":0.30925778,"width":0.070644945,"height":0.030726258},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with Google","depth":9,"bounds":{"left":0.48603722,"top":0.3754988,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with Google","depth":11,"bounds":{"left":0.5398936,"top":0.38028732,"width":0.04454787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with SAML SSO","depth":9,"bounds":{"left":0.48603722,"top":0.40422985,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with SAML SSO","depth":11,"bounds":{"left":0.5362367,"top":0.40901837,"width":0.051861703,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":10,"bounds":{"left":0.55634975,"top":0.44014364,"width":0.0043218085,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Email","depth":10,"bounds":{"left":0.48603722,"top":0.46727854,"width":0.011136968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"you@example.com","depth":10,"bounds":{"left":0.49019283,"top":0.48643255,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"you@example.com","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Password","depth":10,"bounds":{"left":0.48603722,"top":0.5215483,"width":0.020279255,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Forgot password?","depth":10,"bounds":{"left":0.5940825,"top":0.5215483,"width":0.036901597,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"••••••","depth":10,"bounds":{"left":0.49019283,"top":0.54070234,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"Password","role_description":"secure text field","subrole":"AXSecureTextField","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Log in","depth":9,"bounds":{"left":0.48603722,"top":0.57821226,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Log in","depth":11,"bounds":{"left":0.5521942,"top":0.5830008,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-493634218375676853
|
-2831913821102687342
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22658
|
975
|
43
|
2026-05-12T07:17:09.237592+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570229237_m2.jpg...
|
Firefox
|
Content Explorer — Work
|
True
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.02925532,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"J J Jiminny","depth":8,"bounds":{"left":0.46409574,"top":0.058260176,"width":0.03474069,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"J","depth":11,"bounds":{"left":0.46758643,"top":0.07940942,"width":0.0023271276,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jiminny","depth":9,"bounds":{"left":0.47539893,"top":0.06304868,"width":0.016123671,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Search Jiminny ⌘K","depth":9,"bounds":{"left":0.6296542,"top":0.058260176,"width":0.21143617,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Search Jiminny","depth":11,"bounds":{"left":0.6402925,"top":0.06304868,"width":0.031416222,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"⌘","depth":10,"bounds":{"left":0.82629657,"top":0.06344773,"width":0.0039893617,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"K","depth":10,"bounds":{"left":0.8351064,"top":0.06344773,"width":0.0026595744,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4159630415606179326
|
-2831911623153173866
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K...
|
22656
|
NULL
|
NULL
|
NULL
|
|
22657
|
974
|
33
|
2026-05-12T07:17:09.231626+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570229231_m1.jpg...
|
Firefox
|
Content Explorer — Work
|
True
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-7033255905884690172
|
-2831770885664293998
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22656
|
975
|
42
|
2026-05-12T07:17:08.126461+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570228126_m2.jpg...
|
Firefox
|
Content Explorer — Work
|
True
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell
🚨 Risk and Churn Analytics
Implementation
Impl Projects
Trial Opps (Under Review)
Stoyan's clients
Content Explorer
Content Explorer
Tab icon For you
For you
Tab icon Sections
Sections
Tab icon Pages
Pages
Tab icon Archived
Archived
Add new
Add new
Search...
No content to display
The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace
Add new
Add new
Content Explorer
Content Explorer
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Explore Keynote
Last edited
All Active Renewal Opps
2 minutes ago
All Existing Clients
15 minutes ago
My Renewals next 2Qs
17 minutes ago
OD Tasks
17 minutes ago
First 30 days to adoption
20 minutes ago
Explore community templates
Page templates
Section templates
Julia from Planhat
Julia
from Planhat
Close
Improved Task dependencies in Workflows
Improved Task dependencies in Workflows
Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:
Tasks can have multiple dependencies, across phases
Dependent task dates automatically update when a parent task is moved
Dependent tasks are assigned dates when a workflow is applied
Read more in our
docs opens in new tab...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.02925532,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"J J Jiminny","depth":8,"bounds":{"left":0.46409574,"top":0.058260176,"width":0.03474069,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"J","depth":11,"bounds":{"left":0.46758643,"top":0.07940942,"width":0.0023271276,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jiminny","depth":9,"bounds":{"left":0.47539893,"top":0.06304868,"width":0.016123671,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Search Jiminny ⌘K","depth":9,"bounds":{"left":0.6296542,"top":0.058260176,"width":0.21143617,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Search Jiminny","depth":11,"bounds":{"left":0.6402925,"top":0.06304868,"width":0.031416222,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"⌘","depth":10,"bounds":{"left":0.82629657,"top":0.06344773,"width":0.0039893617,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"K","depth":10,"bounds":{"left":0.8351064,"top":0.06344773,"width":0.0026595744,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"L","depth":11,"bounds":{"left":0.9730718,"top":0.06304868,"width":0.002493351,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Lukas","depth":10,"bounds":{"left":0.9808843,"top":0.06304868,"width":0.012300532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.09377494,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.09936153,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Data Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.11931365,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Data Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.12490024,"width":0.027925532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":12,"bounds":{"left":0.46426198,"top":0.14485236,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Calendar","depth":14,"bounds":{"left":0.47623006,"top":0.15043895,"width":0.01861702,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Notifications","depth":12,"bounds":{"left":0.46426198,"top":0.17039107,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Notifications","depth":14,"bounds":{"left":0.47623006,"top":0.17597765,"width":0.026263298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"More","depth":12,"bounds":{"left":0.46426198,"top":0.19592977,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"More","depth":14,"bounds":{"left":0.47623006,"top":0.20151636,"width":0.010638298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sections","depth":13,"bounds":{"left":0.46426198,"top":0.23264167,"width":0.07114362,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":15,"bounds":{"left":0.46559176,"top":0.23782921,"width":0.01662234,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CS Day-to-day","depth":16,"bounds":{"left":0.47822472,"top":0.26456505,"width":0.030751329,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚀 Getting started Guide","depth":18,"bounds":{"left":0.47822472,"top":0.29010376,"width":0.049700797,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🪬 Just CS Data","depth":18,"bounds":{"left":0.47822472,"top":0.3140463,"width":0.032081116,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"👉 Daily Operations","depth":18,"bounds":{"left":0.47822472,"top":0.33798882,"width":0.03956117,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🗓️ Weekly prep","depth":18,"bounds":{"left":0.47822472,"top":0.36193135,"width":0.03125,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🤑 Renewals and Upsell","depth":18,"bounds":{"left":0.47822472,"top":0.3858739,"width":0.048204787,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚨 Risk and Churn Analytics","depth":18,"bounds":{"left":0.47822472,"top":0.40981644,"width":0.05718085,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Implementation","depth":16,"bounds":{"left":0.47822472,"top":0.44972068,"width":0.032579787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Impl Projects","depth":18,"bounds":{"left":0.47822472,"top":0.47525936,"width":0.026761968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Trial Opps (Under Review)","depth":18,"bounds":{"left":0.47822472,"top":0.49920192,"width":0.054022606,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Stoyan's clients","depth":18,"bounds":{"left":0.47822472,"top":0.5231444,"width":0.032579787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":15,"bounds":{"left":0.54355055,"top":0.09537111,"width":0.046043884,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":17,"bounds":{"left":0.55285907,"top":0.0981644,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon For you","depth":16,"bounds":{"left":0.5944149,"top":0.08739027,"width":0.028091755,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"For you","depth":18,"bounds":{"left":0.6043883,"top":0.0981644,"width":0.015458777,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Sections","depth":16,"bounds":{"left":0.62317157,"top":0.08739027,"width":0.030585106,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":18,"bounds":{"left":0.633145,"top":0.0981644,"width":0.017952127,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Pages","depth":16,"bounds":{"left":0.65442157,"top":0.08739027,"width":0.025265958,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pages","depth":18,"bounds":{"left":0.664395,"top":0.0981644,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Archived","depth":16,"bounds":{"left":0.6803524,"top":0.08739027,"width":0.031083776,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Archived","depth":18,"bounds":{"left":0.6903258,"top":0.0981644,"width":0.018450798,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Add new","depth":14,"bounds":{"left":0.84890294,"top":0.09377494,"width":0.036402926,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Add new","depth":16,"bounds":{"left":0.85837764,"top":0.0981644,"width":0.018118352,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"Search...","depth":16,"bounds":{"left":0.55502,"top":0.12809257,"width":0.05086436,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"No content to display","depth":19,"bounds":{"left":0.6924867,"top":0.56264967,"width":0.043882977,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace","depth":19,"bounds":{"left":0.6549202,"top":0.5802075,"width":0.119015954,"height":0.027533919},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Add new","depth":18,"bounds":{"left":0.6961436,"top":0.6209098,"width":0.03656915,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Add new","depth":20,"bounds":{"left":0.7056183,"top":0.6256983,"width":0.018284574,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Content Explorer","depth":18,"bounds":{"left":0.56981385,"top":0.91660017,"width":0.050033245,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":20,"bounds":{"left":0.58111703,"top":0.9261772,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Organizing content: Pages, Sections and Libraries","depth":18,"bounds":{"left":0.62117684,"top":0.91660017,"width":0.118351065,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Organizing content: Pages, Sections and Libraries","depth":20,"bounds":{"left":0.63248,"top":0.9261772,"width":0.10305851,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Organizing content: Pages, Sections and Libraries","depth":18,"bounds":{"left":0.7408577,"top":0.91660017,"width":0.11818484,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Organizing content: Pages, Sections and Libraries","depth":20,"bounds":{"left":0.7521609,"top":0.9261772,"width":0.10289229,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Explore Keynote","depth":15,"bounds":{"left":0.9243683,"top":0.3292099,"width":0.0390625,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Last edited","depth":16,"bounds":{"left":0.89744014,"top":0.36871508,"width":0.021110373,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All Active Renewal Opps","depth":18,"bounds":{"left":0.9054189,"top":0.3934557,"width":0.04637633,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2 minutes ago","depth":18,"bounds":{"left":0.9659242,"top":0.39385474,"width":0.02443484,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All Existing Clients","depth":18,"bounds":{"left":0.9054189,"top":0.41660017,"width":0.03507314,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.4169992,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"My Renewals next 2Qs","depth":18,"bounds":{"left":0.9054189,"top":0.43974462,"width":0.04338431,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.44014364,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OD Tasks","depth":18,"bounds":{"left":0.9054189,"top":0.46288908,"width":0.018118352,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.4632881,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"First 30 days to adoption","depth":18,"bounds":{"left":0.9054189,"top":0.48603353,"width":0.04737367,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20 minutes ago","depth":18,"bounds":{"left":0.96359706,"top":0.48643255,"width":0.026761968,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Explore community templates","depth":16,"bounds":{"left":0.89744014,"top":0.5203512,"width":0.05651596,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Page templates","depth":18,"bounds":{"left":0.90807843,"top":0.5510774,"width":0.029421542,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Section templates","depth":18,"bounds":{"left":0.90807843,"top":0.5798085,"width":0.034242023,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Julia from Planhat","depth":17,"bounds":{"left":0.90176195,"top":0.6927374,"width":0.07247341,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Julia","depth":18,"bounds":{"left":0.90176195,"top":0.6927374,"width":0.00930851,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"from Planhat","depth":19,"bounds":{"left":0.91107047,"top":0.6927374,"width":0.026761968,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close","depth":15,"bounds":{"left":0.96758646,"top":0.6671987,"width":0.023936171,"height":0.057462092},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"Improved Task dependencies in Workflows","depth":17,"bounds":{"left":0.88613695,"top":0.7246608,"width":0.09707447,"height":0.0490822},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Improved Task dependencies in Workflows","depth":18,"bounds":{"left":0.88613695,"top":0.7258579,"width":0.09142287,"height":0.038707104},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:","depth":18,"bounds":{"left":0.88613695,"top":0.7869114,"width":0.094082445,"height":0.060255386},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Tasks can have multiple dependencies, across phases","depth":19,"bounds":{"left":0.8937833,"top":0.85714287,"width":0.08377659,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Dependent task dates automatically update when a parent task is moved","depth":19,"bounds":{"left":0.8937833,"top":0.8910614,"width":0.07762633,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Dependent tasks are assigned dates when a workflow is applied","depth":19,"bounds":{"left":0.8937833,"top":0.92498004,"width":0.07829122,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Read more in our","depth":18,"bounds":{"left":0.88613695,"top":1.0,"width":0.037898935,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"docs opens in new tab","depth":18,"bounds":{"left":0.9240359,"top":1.0,"width":0.010472074,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
288263493835329509
|
-2831700515846899050
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell
🚨 Risk and Churn Analytics
Implementation
Impl Projects
Trial Opps (Under Review)
Stoyan's clients
Content Explorer
Content Explorer
Tab icon For you
For you
Tab icon Sections
Sections
Tab icon Pages
Pages
Tab icon Archived
Archived
Add new
Add new
Search...
No content to display
The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace
Add new
Add new
Content Explorer
Content Explorer
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Explore Keynote
Last edited
All Active Renewal Opps
2 minutes ago
All Existing Clients
15 minutes ago
My Renewals next 2Qs
17 minutes ago
OD Tasks
17 minutes ago
First 30 days to adoption
20 minutes ago
Explore community templates
Page templates
Section templates
Julia from Planhat
Julia
from Planhat
Close
Improved Task dependencies in Workflows
Improved Task dependencies in Workflows
Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:
Tasks can have multiple dependencies, across phases
Dependent task dates automatically update when a parent task is moved
Dependent tasks are assigned dates when a workflow is applied
Read more in our
docs opens in new tab...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22655
|
974
|
32
|
2026-05-12T07:17:08.126456+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570228126_m1.jpg...
|
Firefox
|
Content Explorer — Work
|
True
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.13090278,"top":0.0,"width":0.24027778,"height":0.026666667},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.13090278,"top":0.0,"width":0.18402778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-2318075514972954299
|
-2831770885395858542
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup...
|
22649
|
NULL
|
NULL
|
NULL
|
|
22654
|
975
|
41
|
2026-05-12T07:16:57.209914+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570217209_m2.jpg...
|
Firefox
|
Content Explorer — Work
|
True
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell
🚨 Risk and Churn Analytics
Implementation
Impl Projects
Trial Opps (Under Review)
Stoyan's clients
Content Explorer
Content Explorer
Tab icon For you
For you
Tab icon Sections
Sections
Tab icon Pages
Pages
Tab icon Archived
Archived
Add new
Add new
Search...
No content to display
The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace
Add new
Add new
Content Explorer
Content Explorer
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Explore Keynote
Last edited
All Active Renewal Opps
2 minutes ago
All Existing Clients
15 minutes ago
My Renewals next 2Qs
17 minutes ago
OD Tasks
17 minutes ago
First 30 days to adoption
20 minutes ago
Explore community templates
Page templates
Section templates
Julia from Planhat
Julia
from Planhat
Close
Improved Task dependencies in Workflows
Improved Task dependencies in Workflows
Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:
Tasks can have multiple dependencies, across phases
Dependent task dates automatically update when a parent task is moved
Dependent tasks are assigned dates when a workflow is applied
Read more in our
docs opens in new tab
docs
.
...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.02925532,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"J J Jiminny","depth":8,"bounds":{"left":0.46409574,"top":0.058260176,"width":0.03474069,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"J","depth":11,"bounds":{"left":0.46758643,"top":0.07940942,"width":0.0023271276,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jiminny","depth":9,"bounds":{"left":0.47539893,"top":0.06304868,"width":0.016123671,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Search Jiminny ⌘K","depth":9,"bounds":{"left":0.6296542,"top":0.058260176,"width":0.21143617,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Search Jiminny","depth":11,"bounds":{"left":0.6402925,"top":0.06304868,"width":0.031416222,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"⌘","depth":10,"bounds":{"left":0.82629657,"top":0.06344773,"width":0.0039893617,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"K","depth":10,"bounds":{"left":0.8351064,"top":0.06344773,"width":0.0026595744,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"L","depth":11,"bounds":{"left":0.9730718,"top":0.06304868,"width":0.002493351,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Lukas","depth":10,"bounds":{"left":0.9808843,"top":0.06304868,"width":0.012300532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.09377494,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.09936153,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Data Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.11931365,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Data Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.12490024,"width":0.027925532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":12,"bounds":{"left":0.46426198,"top":0.14485236,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Calendar","depth":14,"bounds":{"left":0.47623006,"top":0.15043895,"width":0.01861702,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Notifications","depth":12,"bounds":{"left":0.46426198,"top":0.17039107,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Notifications","depth":14,"bounds":{"left":0.47623006,"top":0.17597765,"width":0.026263298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"More","depth":12,"bounds":{"left":0.46426198,"top":0.19592977,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"More","depth":14,"bounds":{"left":0.47623006,"top":0.20151636,"width":0.010638298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sections","depth":13,"bounds":{"left":0.46426198,"top":0.23264167,"width":0.07114362,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":15,"bounds":{"left":0.46559176,"top":0.23782921,"width":0.01662234,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CS Day-to-day","depth":16,"bounds":{"left":0.47822472,"top":0.26456505,"width":0.030751329,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚀 Getting started Guide","depth":18,"bounds":{"left":0.47822472,"top":0.29010376,"width":0.049700797,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🪬 Just CS Data","depth":18,"bounds":{"left":0.47822472,"top":0.3140463,"width":0.032081116,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"👉 Daily Operations","depth":18,"bounds":{"left":0.47822472,"top":0.33798882,"width":0.03956117,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🗓️ Weekly prep","depth":18,"bounds":{"left":0.47822472,"top":0.36193135,"width":0.03125,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🤑 Renewals and Upsell","depth":18,"bounds":{"left":0.47822472,"top":0.3858739,"width":0.048204787,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚨 Risk and Churn Analytics","depth":18,"bounds":{"left":0.47822472,"top":0.40981644,"width":0.05718085,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Implementation","depth":16,"bounds":{"left":0.47822472,"top":0.44972068,"width":0.032579787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Impl Projects","depth":18,"bounds":{"left":0.47822472,"top":0.47525936,"width":0.026761968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Trial Opps (Under Review)","depth":18,"bounds":{"left":0.47822472,"top":0.49920192,"width":0.054022606,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Stoyan's clients","depth":18,"bounds":{"left":0.47822472,"top":0.5231444,"width":0.032579787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":15,"bounds":{"left":0.54355055,"top":0.09537111,"width":0.046043884,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":17,"bounds":{"left":0.55285907,"top":0.0981644,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon For you","depth":16,"bounds":{"left":0.5944149,"top":0.08739027,"width":0.028091755,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"For you","depth":18,"bounds":{"left":0.6043883,"top":0.0981644,"width":0.015458777,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Sections","depth":16,"bounds":{"left":0.62317157,"top":0.08739027,"width":0.030585106,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":18,"bounds":{"left":0.633145,"top":0.0981644,"width":0.017952127,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Pages","depth":16,"bounds":{"left":0.65442157,"top":0.08739027,"width":0.025265958,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pages","depth":18,"bounds":{"left":0.664395,"top":0.0981644,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Archived","depth":16,"bounds":{"left":0.6803524,"top":0.08739027,"width":0.031083776,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Archived","depth":18,"bounds":{"left":0.6903258,"top":0.0981644,"width":0.018450798,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Add new","depth":14,"bounds":{"left":0.84890294,"top":0.09377494,"width":0.036402926,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Add new","depth":16,"bounds":{"left":0.85837764,"top":0.0981644,"width":0.018118352,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"Search...","depth":16,"bounds":{"left":0.55502,"top":0.12809257,"width":0.05086436,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"No content to display","depth":19,"bounds":{"left":0.6924867,"top":0.56264967,"width":0.043882977,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace","depth":19,"bounds":{"left":0.6549202,"top":0.5802075,"width":0.119015954,"height":0.027533919},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Add new","depth":18,"bounds":{"left":0.6961436,"top":0.6209098,"width":0.03656915,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Add new","depth":20,"bounds":{"left":0.7056183,"top":0.6256983,"width":0.018284574,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Content Explorer","depth":18,"bounds":{"left":0.56981385,"top":0.91660017,"width":0.050033245,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":20,"bounds":{"left":0.58111703,"top":0.9261772,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Organizing content: Pages, Sections and Libraries","depth":18,"bounds":{"left":0.62117684,"top":0.91660017,"width":0.118351065,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Organizing content: Pages, Sections and Libraries","depth":20,"bounds":{"left":0.63248,"top":0.9261772,"width":0.10305851,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Organizing content: Pages, Sections and Libraries","depth":18,"bounds":{"left":0.7408577,"top":0.91660017,"width":0.11818484,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Organizing content: Pages, Sections and Libraries","depth":20,"bounds":{"left":0.7521609,"top":0.9261772,"width":0.10289229,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Explore Keynote","depth":15,"bounds":{"left":0.9243683,"top":0.3292099,"width":0.0390625,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Last edited","depth":16,"bounds":{"left":0.89744014,"top":0.36871508,"width":0.021110373,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All Active Renewal Opps","depth":18,"bounds":{"left":0.9054189,"top":0.3934557,"width":0.04637633,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2 minutes ago","depth":18,"bounds":{"left":0.9659242,"top":0.39385474,"width":0.02443484,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All Existing Clients","depth":18,"bounds":{"left":0.9054189,"top":0.41660017,"width":0.03507314,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.4169992,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"My Renewals next 2Qs","depth":18,"bounds":{"left":0.9054189,"top":0.43974462,"width":0.04338431,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.44014364,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OD Tasks","depth":18,"bounds":{"left":0.9054189,"top":0.46288908,"width":0.018118352,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.4632881,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"First 30 days to adoption","depth":18,"bounds":{"left":0.9054189,"top":0.48603353,"width":0.04737367,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20 minutes ago","depth":18,"bounds":{"left":0.96359706,"top":0.48643255,"width":0.026761968,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Explore community templates","depth":16,"bounds":{"left":0.89744014,"top":0.5203512,"width":0.05651596,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Page templates","depth":18,"bounds":{"left":0.90807843,"top":0.5510774,"width":0.029421542,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Section templates","depth":18,"bounds":{"left":0.90807843,"top":0.5798085,"width":0.034242023,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Julia from Planhat","depth":17,"bounds":{"left":0.90176195,"top":0.6927374,"width":0.07247341,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Julia","depth":18,"bounds":{"left":0.90176195,"top":0.6927374,"width":0.00930851,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"from Planhat","depth":19,"bounds":{"left":0.91107047,"top":0.6927374,"width":0.026761968,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close","depth":15,"bounds":{"left":0.96758646,"top":0.6671987,"width":0.023936171,"height":0.057462092},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"Improved Task dependencies in Workflows","depth":17,"bounds":{"left":0.88613695,"top":0.7246608,"width":0.09707447,"height":0.0490822},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Improved Task dependencies in Workflows","depth":18,"bounds":{"left":0.88613695,"top":0.7258579,"width":0.09142287,"height":0.038707104},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:","depth":18,"bounds":{"left":0.88613695,"top":0.7869114,"width":0.094082445,"height":0.060255386},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Tasks can have multiple dependencies, across phases","depth":19,"bounds":{"left":0.8937833,"top":0.85714287,"width":0.08377659,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Dependent task dates automatically update when a parent task is moved","depth":19,"bounds":{"left":0.8937833,"top":0.8910614,"width":0.07762633,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Dependent tasks are assigned dates when a workflow is applied","depth":19,"bounds":{"left":0.8937833,"top":0.92498004,"width":0.07829122,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Read more in our","depth":18,"bounds":{"left":0.88613695,"top":1.0,"width":0.037898935,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"docs opens in new tab","depth":18,"bounds":{"left":0.9240359,"top":1.0,"width":0.010472074,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"docs","depth":19,"bounds":{"left":0.9240359,"top":1.0,"width":0.010472074,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":18,"bounds":{"left":0.93450797,"top":1.0,"width":0.0013297872,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"","depth":18,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-1272291732256829089
|
-2831700515846899050
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell
🚨 Risk and Churn Analytics
Implementation
Impl Projects
Trial Opps (Under Review)
Stoyan's clients
Content Explorer
Content Explorer
Tab icon For you
For you
Tab icon Sections
Sections
Tab icon Pages
Pages
Tab icon Archived
Archived
Add new
Add new
Search...
No content to display
The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace
Add new
Add new
Content Explorer
Content Explorer
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Explore Keynote
Last edited
All Active Renewal Opps
2 minutes ago
All Existing Clients
15 minutes ago
My Renewals next 2Qs
17 minutes ago
OD Tasks
17 minutes ago
First 30 days to adoption
20 minutes ago
Explore community templates
Page templates
Section templates
Julia from Planhat
Julia
from Planhat
Close
Improved Task dependencies in Workflows
Improved Task dependencies in Workflows
Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:
Tasks can have multiple dependencies, across phases
Dependent task dates automatically update when a parent task is moved
Dependent tasks are assigned dates when a workflow is applied
Read more in our
docs opens in new tab
docs
.
...
|
22653
|
NULL
|
NULL
|
NULL
|
|
22653
|
975
|
40
|
2026-05-12T07:16:54.162847+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570214162_m2.jpg...
|
Firefox
|
Home — Work
|
True
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Home
Home
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Home","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Home","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.010139627,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"J J Jiminny","depth":8,"bounds":{"left":0.46409574,"top":0.058260176,"width":0.03474069,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"J","depth":11,"bounds":{"left":0.46758643,"top":0.07940942,"width":0.0023271276,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jiminny","depth":9,"bounds":{"left":0.47539893,"top":0.06304868,"width":0.016123671,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Search Jiminny ⌘K","depth":9,"bounds":{"left":0.6296542,"top":0.058260176,"width":0.21143617,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Search Jiminny","depth":11,"bounds":{"left":0.6402925,"top":0.06304868,"width":0.031416222,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"⌘","depth":10,"bounds":{"left":0.82629657,"top":0.06344773,"width":0.0039893617,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"K","depth":10,"bounds":{"left":0.8351064,"top":0.06344773,"width":0.0026595744,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"L","depth":11,"bounds":{"left":0.9730718,"top":0.06304868,"width":0.002493351,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Lukas","depth":10,"bounds":{"left":0.9808843,"top":0.06304868,"width":0.012300532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.09377494,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.09936153,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Data Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.11931365,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Data Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.12490024,"width":0.027925532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":12,"bounds":{"left":0.46426198,"top":0.14485236,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Calendar","depth":14,"bounds":{"left":0.47623006,"top":0.15043895,"width":0.01861702,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Notifications","depth":12,"bounds":{"left":0.46426198,"top":0.17039107,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Notifications","depth":14,"bounds":{"left":0.47623006,"top":0.17597765,"width":0.026263298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"More","depth":12,"bounds":{"left":0.46426198,"top":0.19592977,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"More","depth":14,"bounds":{"left":0.47623006,"top":0.20151636,"width":0.010638298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sections","depth":13,"bounds":{"left":0.46426198,"top":0.23264167,"width":0.07114362,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":15,"bounds":{"left":0.46559176,"top":0.23782921,"width":0.01662234,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CS Day-to-day","depth":16,"bounds":{"left":0.47822472,"top":0.26456505,"width":0.030751329,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚀 Getting started Guide","depth":18,"bounds":{"left":0.47822472,"top":0.29010376,"width":0.049700797,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🪬 Just CS Data","depth":18,"bounds":{"left":0.47822472,"top":0.3140463,"width":0.032081116,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"👉 Daily Operations","depth":18,"bounds":{"left":0.47822472,"top":0.33798882,"width":0.03956117,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🗓️ Weekly prep","depth":18,"bounds":{"left":0.47822472,"top":0.36193135,"width":0.03125,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🤑 Renewals and Upsell","depth":18,"bounds":{"left":0.47822472,"top":0.3858739,"width":0.048204787,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-2672622827034840214
|
-2831911622079432046
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Home
Home
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22652
|
975
|
39
|
2026-05-12T07:16:51.138262+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570211138_m2.jpg...
|
Firefox
|
Planhat — Work
|
True
|
ws.planhat.com/jiminny
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
5710214587205052025
|
-2831913822176429418
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page...
|
22651
|
NULL
|
NULL
|
NULL
|
|
22651
|
975
|
38
|
2026-05-12T07:16:48.128581+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570208128_m2.jpg...
|
Firefox
|
Planhat — Work
|
True
|
ws.planhat.com/login?ga=1
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Read apis.google.com...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Read apis.google.com","depth":5,"bounds":{"left":0.46010637,"top":0.9876297,"width":0.038397606,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-5280693668421536418
|
-2831913822176429418
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Read apis.google.com...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22649
|
974
|
31
|
2026-05-12T07:16:46.046363+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570206046_m1.jpg...
|
Firefox
|
Sign in - Google Accounts — Work
|
True
|
accounts.google.com/v3/signin/accountchooser?clien accounts.google.com/v3/signin/accountchooser?client_id=1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com&redirect_uri=https%3A%2F%2Fapi.planhat.com%2Foauth2%2Fgoogle&response_type=code&scope=email&state=backlink%3D%2F%26ws%3Dtrue%26nonce%3D98e5dc1d37a377b73ed53f4504cc0cc0&dsh=S-1608494723%3A1778570203972862&o2v=2&service=lso&flowName=GeneralOAuthFlow&opparams=%253F&continue=https%3A%2F%2Faccounts.google.com%2Fsignin%2Foauth%2Fconsent%3Fauthuser%3Dunknown%26part%3DAJi8hAMhmYbVUs8yZK4zHz4UXpqvL3xQ9fvdfnlb5j7FPEpY2VxXwhDsnTorT-HT90Uq10gkYTnW_W_VBLZbi-cHAF6mVP6ROLrmd8JIWvHqfO6R5R-ec8qzfpYKmEhVeMqTakdQlVCokOwUHJYC90GdmGeGN5aDrfkpRUAmBVW4uyTGHcZo58zLBpGw-AMuwZS1Zoj07Kqqw3SHF3d3gfPMJAShUQjaX3j7kyKd_wMWc9saMsvGv-gCka7anrItpDc_Rvy-0stPRsJwkPRzkkj-c8LlbdhRiWlsFAym2tgaWPNRlwQQJibrJgku9KNh9soO3lJQ7SfY-d2tJ-ucRpUYIRT4kELx1hsOPdY0iOAh28FsnKWzH4yLVuB5s81qUa2Dzyl53DYXM7q0quaQDT-uCvkQylhfKTzIZ2nLbF29abYaOrMCvPwqMeQ74haMpGm0WNtBFxKptfKyiZBiDQeBNYRiJGSIpg%26flowName%3DGeneralOAuthFlow%26as%3DS-1608494723%253A1778570203972862%26client_id%3D1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com%26requestPath%3D%252Fsignin%252Foauth%252Fconsent%23&app_domain=https%3A%2F%2Fapi.planhat.com...
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores)....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Sign in - Google Accounts","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Sign in - Google Accounts","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-5097155351300601237
|
-2831770884053681262
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores)....
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22650
|
975
|
37
|
2026-05-12T07:16:46.029359+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570206029_m2.jpg...
|
Firefox
|
Sign in - Google Accounts — Work
|
True
|
accounts.google.com/v3/signin/accountchooser?clien accounts.google.com/v3/signin/accountchooser?client_id=1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com&redirect_uri=https%3A%2F%2Fapi.planhat.com%2Foauth2%2Fgoogle&response_type=code&scope=email&state=backlink%3D%2F%26ws%3Dtrue%26nonce%3D98e5dc1d37a377b73ed53f4504cc0cc0&dsh=S-1608494723%3A1778570203972862&o2v=2&service=lso&flowName=GeneralOAuthFlow&opparams=%253F&continue=https%3A%2F%2Faccounts.google.com%2Fsignin%2Foauth%2Fconsent%3Fauthuser%3Dunknown%26part%3DAJi8hAMhmYbVUs8yZK4zHz4UXpqvL3xQ9fvdfnlb5j7FPEpY2VxXwhDsnTorT-HT90Uq10gkYTnW_W_VBLZbi-cHAF6mVP6ROLrmd8JIWvHqfO6R5R-ec8qzfpYKmEhVeMqTakdQlVCokOwUHJYC90GdmGeGN5aDrfkpRUAmBVW4uyTGHcZo58zLBpGw-AMuwZS1Zoj07Kqqw3SHF3d3gfPMJAShUQjaX3j7kyKd_wMWc9saMsvGv-gCka7anrItpDc_Rvy-0stPRsJwkPRzkkj-c8LlbdhRiWlsFAym2tgaWPNRlwQQJibrJgku9KNh9soO3lJQ7SfY-d2tJ-ucRpUYIRT4kELx1hsOPdY0iOAh28FsnKWzH4yLVuB5s81qUa2Dzyl53DYXM7q0quaQDT-uCvkQylhfKTzIZ2nLbF29abYaOrMCvPwqMeQ74haMpGm0WNtBFxKptfKyiZBiDQeBNYRiJGSIpg%26flowName%3DGeneralOAuthFlow%26as%3DS-1608494723%253A1778570203972862%26client_id%3D1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com%26requestPath%3D%252Fsignin%252Foauth%252Fconsent%23&app_domain=https%3A%2F%2Fapi.planhat.com...
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Sign in - Google Accounts","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Sign in - Google Accounts","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.045212764,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
6514073522262166208
|
-2831911622079432046
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page...
|
22648
|
NULL
|
NULL
|
NULL
|
|
22648
|
975
|
36
|
2026-05-12T07:16:45.086821+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570205086_m2.jpg...
|
Firefox
|
Sign in - Google Accounts — Work
|
True
|
accounts.google.com/v3/signin/accountchooser?clien accounts.google.com/v3/signin/accountchooser?client_id=1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com&redirect_uri=https%3A%2F%2Fapi.planhat.com%2Foauth2%2Fgoogle&response_type=code&scope=email&state=backlink%3D%2F%26ws%3Dtrue%26nonce%3D98e5dc1d37a377b73ed53f4504cc0cc0&dsh=S-1608494723%3A1778570203972862&o2v=2&service=lso&flowName=GeneralOAuthFlow&opparams=%253F&continue=https%3A%2F%2Faccounts.google.com%2Fsignin%2Foauth%2Fconsent%3Fauthuser%3Dunknown%26part%3DAJi8hAMhmYbVUs8yZK4zHz4UXpqvL3xQ9fvdfnlb5j7FPEpY2VxXwhDsnTorT-HT90Uq10gkYTnW_W_VBLZbi-cHAF6mVP6ROLrmd8JIWvHqfO6R5R-ec8qzfpYKmEhVeMqTakdQlVCokOwUHJYC90GdmGeGN5aDrfkpRUAmBVW4uyTGHcZo58zLBpGw-AMuwZS1Zoj07Kqqw3SHF3d3gfPMJAShUQjaX3j7kyKd_wMWc9saMsvGv-gCka7anrItpDc_Rvy-0stPRsJwkPRzkkj-c8LlbdhRiWlsFAym2tgaWPNRlwQQJibrJgku9KNh9soO3lJQ7SfY-d2tJ-ucRpUYIRT4kELx1hsOPdY0iOAh28FsnKWzH4yLVuB5s81qUa2Dzyl53DYXM7q0quaQDT-uCvkQylhfKTzIZ2nLbF29abYaOrMCvPwqMeQ74haMpGm0WNtBFxKptfKyiZBiDQeBNYRiJGSIpg%26flowName%3DGeneralOAuthFlow%26as%3DS-1608494723%253A1778570203972862%26client_id%3D1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com%26requestPath%3D%252Fsignin%252Foauth%252Fconsent%23&app_domain=https%3A%2F%2Fapi.planhat.com...
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Sign in with Google
Choose an account
Choose an account
to continue to
Planhat
[EMAIL]
[EMAIL]
Integration Account [EMAIL] Signed out
Integration Account
[EMAIL]
Signed out
Use another account
Use another account
Before using this app, you can review Planhat’s
Privacy Policy
Privacy Policy
and
Terms of Service
Terms of Service
.
English (United States)
English (United States)
Help...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Sign in - Google Accounts","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Sign in - Google Accounts","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.045212764,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Sign in with Google","depth":11,"bounds":{"left":0.5724734,"top":0.35434955,"width":0.042054523,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Choose an account","depth":10,"bounds":{"left":0.56848407,"top":0.4640862,"width":0.15292554,"height":0.0415004},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Choose an account","depth":11,"bounds":{"left":0.56848407,"top":0.46249002,"width":0.12832446,"height":0.044692736},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to continue to","depth":11,"bounds":{"left":0.56848407,"top":0.5199521,"width":0.03474069,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Planhat","depth":11,"bounds":{"left":0.60322475,"top":0.5199521,"width":0.019614361,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXLink","text":"lukas.kovalik@jiminny.com","depth":14,"bounds":{"left":0.73204786,"top":0.42577812,"width":0.16356383,"height":0.0415004},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"lukas.kovalik@jiminny.com","depth":16,"bounds":{"left":0.7506649,"top":0.43934557,"width":0.0546875,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Integration Account integration-account@jiminny.com Signed out","depth":14,"bounds":{"left":0.73204786,"top":0.46807662,"width":0.16356383,"height":0.054269753},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Integration Account","depth":16,"bounds":{"left":0.7506649,"top":0.4792498,"width":0.04936835,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"integration-account@jiminny.com","depth":16,"bounds":{"left":0.7506649,"top":0.49760574,"width":0.070644945,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Signed out","depth":16,"bounds":{"left":0.8703458,"top":0.47805268,"width":0.019946808,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Use another account","depth":14,"bounds":{"left":0.73204786,"top":0.5231444,"width":0.16356383,"height":0.0415004},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Use another account","depth":16,"bounds":{"left":0.7506649,"top":0.5359138,"width":0.051861703,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Before using this app, you can review Planhat’s","depth":10,"bounds":{"left":0.73736703,"top":0.5989625,"width":0.099734046,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Privacy Policy","depth":10,"bounds":{"left":0.83710104,"top":0.5989625,"width":0.03158245,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Privacy Policy","depth":11,"bounds":{"left":0.83710104,"top":0.5989625,"width":0.03158245,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and","depth":10,"bounds":{"left":0.8686835,"top":0.5989625,"width":0.009142287,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Terms of Service","depth":10,"bounds":{"left":0.73736703,"top":0.6149242,"width":0.03756649,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Terms of Service","depth":11,"bounds":{"left":0.73736703,"top":0.6149242,"width":0.03756649,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":10,"bounds":{"left":0.7749335,"top":0.6149242,"width":0.0013297872,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXComboBox","text":"English (United States)","depth":10,"bounds":{"left":0.55651593,"top":0.67398244,"width":0.061502658,"height":0.025538707},"on_screen":true,"value":"English (United States)","help_text":"","role_description":"combo box","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"English (United States)","depth":12,"bounds":{"left":0.5618351,"top":0.68076617,"width":0.04155585,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Help","depth":11,"bounds":{"left":0.8387633,"top":0.6675978,"width":0.016456118,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
8855053595188441098
|
-2831841528213161322
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Sign in with Google
Choose an account
Choose an account
to continue to
Planhat
[EMAIL]
[EMAIL]
Integration Account [EMAIL] Signed out
Integration Account
[EMAIL]
Signed out
Use another account
Use another account
Before using this app, you can review Planhat’s
Privacy Policy
Privacy Policy
and
Terms of Service
Terms of Service
.
English (United States)
English (United States)
Help...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22646
|
974
|
30
|
2026-05-12T07:16:43.376911+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570203376_m1.jpg...
|
Firefox
|
Login — Work
|
True
|
ws.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.13090278,"top":0.0,"width":0.24027778,"height":0.026666667},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.13090278,"top":0.0,"width":0.18402778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.09652778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.19861111,"height":0.10944445},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.09861111,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.20173611,"height":0.10944445},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"bounds":{"left":0.16527778,"top":0.19944444,"width":0.017708333,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"bounds":{"left":0.18715277,"top":0.19777778,"width":0.11215278,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"bounds":{"left":0.123958334,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"bounds":{"left":0.14618056,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"bounds":{"left":0.16840278,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"bounds":{"left":0.190625,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"bounds":{"left":0.21284722,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.15729167,"top":0.33444443,"width":0.027777778,"height":0.044444446},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"bounds":{"left":0.18784723,"top":0.33444443,"width":0.027777778,"height":0.044444446},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
8733677560677435636
|
-2831911622884213870
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit...
|
22645
|
NULL
|
NULL
|
NULL
|
|
22647
|
975
|
35
|
2026-05-12T07:16:43.349644+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570203349_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
ws.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google Loading
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Welcome back","depth":9,"bounds":{"left":0.48603722,"top":0.30526736,"width":0.14494681,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Welcome back","depth":10,"bounds":{"left":0.48603722,"top":0.30925778,"width":0.070644945,"height":0.030726258},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with Google Loading","depth":9,"bounds":{"left":0.48603722,"top":0.3754988,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with Google","depth":11,"bounds":{"left":0.5398936,"top":0.38028732,"width":0.04454787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with SAML SSO","depth":9,"bounds":{"left":0.48603722,"top":0.40422985,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with SAML SSO","depth":11,"bounds":{"left":0.5362367,"top":0.40901837,"width":0.051861703,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":10,"bounds":{"left":0.55634975,"top":0.44014364,"width":0.0043218085,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Email","depth":10,"bounds":{"left":0.48603722,"top":0.46727854,"width":0.011136968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"you@example.com","depth":10,"bounds":{"left":0.49019283,"top":0.48643255,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"you@example.com","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Password","depth":10,"bounds":{"left":0.48603722,"top":0.5215483,"width":0.020279255,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Forgot password?","depth":10,"bounds":{"left":0.5940825,"top":0.5215483,"width":0.036901597,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"••••••","depth":10,"bounds":{"left":0.49019283,"top":0.54070234,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"Password","role_description":"secure text field","subrole":"AXSecureTextField","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Log in","depth":9,"bounds":{"left":0.48603722,"top":0.57821226,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Log in","depth":11,"bounds":{"left":0.5521942,"top":0.5830008,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
6791598485010019321
|
-2831913821102687342
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google Loading
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
22644
|
NULL
|
NULL
|
NULL
|
|
22645
|
974
|
29
|
2026-05-12T07:16:40.906544+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570200906_m1.jpg...
|
Firefox
|
Login — Work
|
True
|
ws.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-4925565648599463165
|
-2831911623152649326
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22644
|
975
|
34
|
2026-05-12T07:16:36.020780+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570196020_m2.jpg...
|
Firefox
|
Login — Work
|
True
|
ws.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Welcome back","depth":9,"bounds":{"left":0.48603722,"top":0.30526736,"width":0.14494681,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Welcome back","depth":10,"bounds":{"left":0.48603722,"top":0.30925778,"width":0.070644945,"height":0.030726258},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with Google","depth":9,"bounds":{"left":0.48603722,"top":0.3754988,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with Google","depth":11,"bounds":{"left":0.5398936,"top":0.38028732,"width":0.04454787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with SAML SSO","depth":9,"bounds":{"left":0.48603722,"top":0.40422985,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with SAML SSO","depth":11,"bounds":{"left":0.5362367,"top":0.40901837,"width":0.051861703,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":10,"bounds":{"left":0.55634975,"top":0.44014364,"width":0.0043218085,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Email","depth":10,"bounds":{"left":0.48603722,"top":0.46727854,"width":0.011136968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"you@example.com","depth":10,"bounds":{"left":0.49019283,"top":0.48643255,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"you@example.com","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Password","depth":10,"bounds":{"left":0.48603722,"top":0.5215483,"width":0.020279255,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Forgot password?","depth":10,"bounds":{"left":0.5940825,"top":0.5215483,"width":0.036901597,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"••••••","depth":10,"bounds":{"left":0.49019283,"top":0.54070234,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"Password","role_description":"secure text field","subrole":"AXSecureTextField","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Log in","depth":9,"bounds":{"left":0.48603722,"top":0.57821226,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Log in","depth":11,"bounds":{"left":0.5521942,"top":0.5830008,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-493634218375676853
|
-2831913821102687342
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22643
|
975
|
33
|
2026-05-12T07:16:33.030084+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570193030_m2.jpg...
|
Firefox
|
Planhat — Work
|
True
|
ws.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
2575318264314049800
|
-2434949366096627058
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini...
|
22642
|
NULL
|
NULL
|
NULL
|
|
22642
|
975
|
32
|
2026-05-12T07:16:30.226903+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570190226_m2.jpg...
|
Firefox
|
Planhat — Work
|
True
|
ws.planhat.com
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false}]...
|
-2868272289771096989
|
-2443956565351368050
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22640
|
975
|
31
|
2026-05-12T07:16:28.634517+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570188634_m2.jpg...
|
Firefox
|
Planhat — Work
|
True
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
9052336404017279817
|
-4245703323335019890
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira...
|
22639
|
NULL
|
NULL
|
NULL
|
|
22641
|
974
|
28
|
2026-05-12T07:16:28.609952+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570188609_m1.jpg...
|
Firefox
|
Planhat — Work
|
True
|
www.planhat.com
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
4811382151791849269
|
4457142791016475577
|
click
|
hybrid
|
NULL
|
New Tab
QuickTime PlayerFileEditViewWindowHelpPetk New Tab
QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 44 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:16:28+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
22638
|
NULL
|
NULL
|
NULL
|
|
22638
|
974
|
27
|
2026-05-12T07:16:27.041738+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570187041_m1.jpg...
|
Firefox
|
Planhat — Work
|
True
|
www.planhat.com
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-2517039603172596306
|
-3380775110537792864
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app...
|
NULL
|
NULL
|
NULL
|
NULL
|