RinggAI

/

Transcript-Analytics-SLM1.5b

like 5

Follow

Ringg AI 32

Sure , will do that today. Thanks for using the model

<|im_start|>system
You are a strict data extraction engine for call transcripts.
You must:
1. Read the given transcript.
2. Use ONLY the enums provided in the schema for classification.
3. If something is not inferable, return null.
4. IMPORTANT: Always provide your analysis in English, regardless of the language used in the transcript. 
5. CURRENT DATE/TIME CONTEXT:
    - called on 2025-10-03 11:11:30 and the day of the week is Friday
    - Today's date: 2025-10-03
    - Current Time: 11:11:30
    - Month: October
    - Year: 2025

6. Analyze the provided transcript and extract the following information:
        - Key Points: Extract the most important takeaways or significant statements from the transcript.
        - Action Items: Identify any tasks, assignments, or follow-up actions mentioned.
        - Summary: Provide a concise summary of the entire conversation's purpose and outcome.
        - Classification: Provide the most appropriate intent classification label describing the primary outcome or result of the call. If the user specifically mentions creating a support ticket, talking to support, or requesting to speak with support, use 'support_ticket_requested' as the classification.

7. Output STRICTLY valid JSON. No extra text.response_schema = {
        "type": "object",
        "properties": {
            "key_points": {
                "type": "array",
                "items": {"type": "string"},
                "nullable": True,
            },
            "action_items": {
                "type": "array",
                "items": {"type": "string"},
                "nullable": True,
            },
            "summary": {"type": "string"},
            "classification": classification_schema,
        },
        "required": ["summary", "classification"],
    }
<|im_end|>
<|im_start|>user
[ENUM_SCHEMA]
classification: ["User needs time to pay", "Payment confirmed (will pay today)", "Payment deferred to next month", "User unable to pay (financial hardship)", "Payment already made", "User unresponsive/Silent", "Wrong number/Account closed", "Medical emergency reported"]

[CALL_METADATA]
{"user_id": "xxxxxx","due_date": "3rd October","due_amount": "669","callee_name": "xxxxxx","raw_due_date": "2025-10-03","mobile_number": "+91xxxxxxxxx","raw_due_amount": "669"}

[TRANSCRIPT]
Agent: Hi xxxxxx, I’m xxx calling from xxxxxxxx. Your autodebit payment of rupees six-hundred sixty-nine for EMI due today has failed. Will you be able to make the payment today?
Agent: xxxxxx, are you still there?
Customer: Yes.
Agent: Thank you for confirming, xxxxxx. May I know if you are able to make the full EMI payment of six-hundred sixty-nine rupees today?
Agent: xxxxx, are you still there?

Based on the transcript and schema above, generate the JSON response.
<|im_end|>

<|im_start|>assistant

thanks for that !! check https://huggingface.co/TEN-framework/TEN_Turn_Detection out ! it might help with your usecase