What are best practices for AI agent voice and call settings?

Last updated: November 5, 2026

Voice and call settings control how an AI agent sounds, listens, handles interruptions, detects voicemail, and processes speech during calls. Configure these settings carefully to make call behavior more natural and reliable.

Before you start

  • Confirm the agent will be assigned to a phone channel.

  • Decide whether the agent needs a single language or multiple languages.

  • Prepare important names, products, or technical terms that may need boosted keywords or pronunciation guidance.

Steps

  1. Open Agents.

  2. Select the agent you want to configure.

  3. Go to the Editor tab.

  4. Review the agent instructions and add clear call-specific guidance.

  5. Open the agent settings for voice and call behavior.

  6. Choose speech-to-text mode based on whether speed or accuracy matters more.

  7. Add boosted keywords for domain-specific vocabulary.

  8. Add pronunciation guide entries for words the agent mispronounces.

  9. Adjust responsiveness and interruption sensitivity.

  10. Configure voicemail behavior.

  11. Save your changes.

  12. Test the agent with phone preview or a controlled call.

Tips

  • Use accuracy-focused transcription for detailed or high-stakes calls.

  • Use boosted keywords for names, locations, product terms, and industry vocabulary.

  • Adjust voice speed and responsiveness in small increments.

  • Test pronunciation changes individually.

  • Use static voicemail messages when a consistent script is required.

Troubleshooting

  • The agent speaks too fast: Lower the voice speed.

  • The agent misses caller interruptions: Increase interruption sensitivity.

  • Transcription errors occur: Add boosted keywords or use an accuracy-focused STT mode.

  • Voicemail behavior is incorrect: Review the voicemail action and message configuration.