The conversation transcript audit feature allows users to gain deeper insights into AI assistant interactions. This powerful tool reveals key events and decision points that occur during conversations, helping you understand why and how your AI assistant responds.
Enabling Audit Mode
To access audit information:
Navigate to the Conversations tab in your Frontline dashboard
Locate the "Audit" toggle switch in the top right corner
Click the toggle to enable audit mode
What Audit Mode Reveals
With audit mode enabled, you'll see additional details alongside the standard conversation transcript, including:
Intent matching: See which intents were triggered by user messages
Conversational flows: Track when the AI enters or exits specific conversation flows
API calls: View details of any external API requests made during the conversation
Response insights:
Response sources (for retrieval-augmented generation)
AI model used for responses
High-level instructions provided to the AI
Benefits of Audit Mode
Troubleshoot unexpected AI behaviors
Optimize conversation flows and intent matching
Ensure proper use of external data sources
Verify adherence to AI guidelines and instructions
Best Practices
Use audit mode during AI assistant development and testing phases
Regularly review audited transcripts to identify areas for improvement after deployment
Combine audit insights with user feedback for a comprehensive view of AI performance
Remember, audit mode is a powerful tool for users. Be mindful of user privacy considerations when reviewing detailed conversation data.