Skip to content

Actions and voices

Actions and Voices are Build surfaces that give agents external capabilities and audio identity.

Integrations are configured from the agent Actions tab. Use Actions when your agent needs to book a meeting, look something up, send a message, update a user profile, or run post-session work.

The current Actions flow has provider-backed integrations when a provider catalog is available and custom actions when you need to connect your own service. Keep secrets in provider credentials or secure backend systems; do not put secret values in prompts, browser code, or public snippets.

After an action exists in the workspace, enable it for the agent from the same Actions tab.

The Voices page lets you browse voices from the managed stack. Voices from configured bring-your-own TTS providers appear automatically.

Current voice tabs are:

  • All voices
  • Saved

Use the filters to search by voice name, gender, accent, tag, or language, then save voices you want to revisit. If voice catalog data does not load, check provider configuration, workspace entitlement, and network or service availability.

Confirm the agent has the required provider configuration and listening behavior before connecting phone, widget voice mode, or WhatsApp voice-adjacent workflows.