Single or Multi-Turn? What Users Really Need—and What’s Happening Inside AI Agent Conversations
Summary:
This is part of a series of Quick Reference Guide "Frequently Asked Questions and Answers"
Content (please ensure you mask any confidential information):
Real end-user insight lives deeper—at the turn level, where every user input and agent response unfolds.
1. To inspect a turn:
• Start with USER_INPUT
• Review LLM_OUTPUT
• Check TURN_NUMBER and session context
• Review SESSION_DURATION_SECONDS to assess overall completion speed
• Use token and execution metrics for performance insight
2. To understand the interaction, determine if it’s a one-off response or a multi-step conversation, and assess whether the user’s goal is achieved efficiently.
2
