Deployment

Voice agents have a dedicated Deployment screen (formerly called Integration). It has three tabs, each suited to a different surface:

• JS SDK — embed the voice agent in your own application code.
• Embed Code — a copy-paste widget for web pages.
• Voice Calling — connect a phone number via a telephony provider.

The standard API and Webhook tabs that appear on text agents are not present on voice agents — voice surfaces are different.

JS SDK

Use the JS SDK to integrate a voice agent into your own web or mobile app code.

The tab provides:

• Installation instructions — how to add the SDK to your project.
• Initialization code — a snippet to instantiate the SDK with your agent and API key.
• Example voice session — a minimal end-to-end code example that starts a voice session.
• API key management — copy or regenerate the API key used by the SDK.

The JS SDK is the right choice when you need fine control over the voice session lifecycle, UI, and event handling.

Embed Code

Use Embed Code to add a voice widget to a web page with a copy-paste snippet.

You can customize before copying:

• Widget size
• Widget position
• Primary color

A live preview panel shows the widget with your customizations applied. The preview starts in voice mode by default — what your users will experience.

Embed is the right choice for getting a voice agent on a marketing site, support portal, or in-product help surface quickly, without writing application code.

Voice Calling

Use Voice Calling to connect the voice agent to a phone number via a third-party telephony provider. See Voice Calling (telephony) for the full setup — inbound number assignment, outbound dialing API, and provider-specific call settings.

Versioning and the Deployment tab

The Deployment tab acts on the currently published version of the voice agent. You must have at least one published version before the SDK key, embed snippet, and voice calling configuration are usable.

See Versioning for the shared versioning model.