Looking for a live coach example

I.would.like an example

here is a checklist example.

AI Prompt:
Analyze the live transcript of a conversation between an agent and a client. Evaluate whether the agent performed the following actions based on their dialogue:

Asked for the client’s name

Asked for the model number of the product

Offered an upgraded warranty

Confirmed the caller didn’t need further assistance

Instructions:

Evaluate the transcript in real-time, considering only the dialogue up to the current point.

If the transcript is incomplete or lacks context for an item, mark it as uncompleted (false) unless previously marked as completed.

Focus only on the agent’s dialogue when assessing whether an action was performed.

Ignore irrelevant dialogue or client responses unless they directly confirm the agent’s action.

Maintain a persistent state for each action: once an action is marked as completed (true), it must remain true and never revert to false, even if no further evidence of the action appears in the transcript.

Only update the JSON output when an action changes from false to true. If no actions change to true, do not update the JSON output.

Return the output as a JSON object with boolean fields for each action: action1_completed, action2_completed, action3_completed, action4_completed, where true indicates the action was performed and false indicates it has not yet been performed.

Output Format: Return a JSON object in the following format only when an action changes from false to true:

{
“action1_completed”: boolean,
“action2_completed”: boolean,
“action3_completed”: boolean,
“action4_completed”: boolean
}

Example:

{
“action1_completed”: true,
“action2_completed”: false,
“action3_completed”: true,
“action4_completed”: false
}

Ensure each action is evaluated independently and accurately based on the agent’s dialogue, and only update the JSON when a new action is completed. Once an action is completed do not change the JSON result. Only return exactly the format as shown do not include the term json or quotation marks

Adaptive Card:

{
“type”: “AdaptiveCard”,
“version”: “1.5”,
“body”: [
{
“type”: “TextBlock”,
“text”: “Agent Action Checklist”,
“weight”: “bolder”,
“size”: “medium”,
“wrap”: true
},
{
“type”: “TextBlock”,
“text”: “${if(action1_completed, ‘:white_check_mark:’, ‘:cross_mark:’)} Asked for the client’s name”,
“wrap”: true,
“spacing”: “small”
},
{
“type”: “TextBlock”,
“text”: “${if(action2_completed, ‘:white_check_mark:’, ‘:cross_mark:’)} Asked for the model number of the product”,
“wrap”: true,
“spacing”: “small”
},
{
“type”: “TextBlock”,
“text”: “${if(action3_completed, ‘:white_check_mark:’, ‘:cross_mark:’)} Offered an upgraded warranty”,
“wrap”: true,
“spacing”: “small”
},
{
“type”: “TextBlock”,
“text”: “${if(action4_completed, ‘:white_check_mark:’, ‘:cross_mark:’)} Confirmed the caller didn’t need further assistance”,
“wrap”: true,
“spacing”: “small”
}
],
“$schema”: “http://adaptivecards.io/schemas/adaptive-card.json
}