Learning Timeline
Key Insights
The Benefits of Anonymity Options
Giving users the choice to remain anonymous at the start of a Voice AI conversation will improve the quality and honesty of the feedback provided.
Actionable Insights
Use LLM to identify specific action items. For example, if a customer mentions another community member's name, the LLM can suggest that the team connect the two directly.
Prompts
LLM Feedback Summarization Prompt
Target:
ChatGPT / GPT-4
You are a customer success assistant. Analyze the following transcript from a voice AI feedback call. Extract the following information:
1. Customer Name
2. Key sentiment (Positive/Negative/Neutral)
3. Main feedback points
4. Specific requests or people mentioned
5. Recommended next action.
Format the output as a concise brief for a Slack notification.
Transcript: [Insert Transcript Variable Here]
Step by Step
Building a Voice AI Feedback Automation Workflow to Slack
- Set up a Voice AI agent (e.g., Vapi, Retell AI, or Bland AI) and name it (e.g., Anna).
- Configure the 'System Prompt' for the agent so it understands that the objective is to collect community feedback.
- Design a Conversation Flow that includes opening questions about privacy/anonymity, followed by open-ended questions about the user experience.
- Set an 'End of Call' trigger within the Voice AI platform to send the conversation transcript to an automation tool (such as Make.com or Zapier).
- Connect an LLM module (GPT-4 or Claude) after the trigger to process the transcript text.
- Input a specific prompt into the LLM module to handle summarization and extract key points (such as names, issues, and suggestions).
- Integrate the Slack 'Send Message' module and select the destination channel (e.g., #notifications).
- Map the summary output from the LLM into the Slack message field for automatic delivery.
- Perform a test call to ensure the data flows correctly from voice -> text -> LLM summary -> Slack notification.
