Skip to main content

Managing AI Agent Calls

Learn how to review, analyze, and manage calls handled by your AI agents.

Maria Cairns avatar
Written by Maria Cairns
Updated over a week ago

Why it matters

Call management helps you understand how an AI agent performs during live phone interactions. Reviewing call logs, transcripts, audio recordings, and extracted data ensures the agent behaves as intended and meets your workflow requirements. Although AI agents can also message people, this guide focuses on monitoring and improving call performance.

Key Concepts

Call log: A list of all calls completed by the agent, including timestamps, contacts, duration, and sentiment.

Call analysis: A detailed view containing the transcript, audio recording, and call metadata.

View data: Structured fields extracted from the call based on your parsing configuration.

Agent version used: Shows which version handled the call, helping you track changes across updates.

AI search on call: A tool that lets you ask questions about the call and instantly summarize or clarify what occurred.

Step-by-Step: Manage AI Agent Calls

  1. Open your agent and select the Calls tab.

  2. Browse the list of calls sorted by date, time, and contact information.

  3. Click View call on any entry to open the detailed analysis.

  4. Review the transcript to understand the flow of the conversation.

  5. Play the call recording using the playback controls at the bottom.

  6. Check the call’s summary and sentiment in the Call details section.

  7. Open View data to see extracted fields defined in your Data parsing settings.

  8. Use the AI search bar to ask questions about the call, such as summaries or agreements.

  9. Clear the search chat to reset the analysis panel.

  10. Return to the Calls tab to continue reviewing additional call records.

Tips and Best Practices

  • Use transcripts to verify that agent instructions and call-specific behaviors are being followed.

  • Compare recordings with the transcripts to assess transcription accuracy.

  • Check the agent version for each call to understand whether behavior changes align with recent edits.

  • If data extraction looks incorrect, revisit Data parsing configuration and refine examples.

  • Review calls regularly to identify issues early and improve instructions or call settings.

  • Use AI search to quickly understand long or complex calls.

Troubleshooting

Issue

Possible Cause

Fix

Missing call transcript

Audio not processed or STT configuration issue

Review speech-to-text settings and denoising options

Extracted data is inaccurate

Parsing instructions unclear or missing examples

Improve parsing field descriptions and add example formats

Call recording does not play

Audio file not generated or corrupted

Retry the call preview and check behavior on subsequent calls

Version shown does not match expectations

Agent was edited after the call occurred

Review version history to confirm when changes were made

Did this answer your question?