Native Speech-to-Speech Voice AI Support in Chat Widget

1. S2S and existing Voice AI agents are available within the

   same Agent dropdown

   .

2. The Chat Widget automatically identifies the selected agent type and launches the appropriate voice experience.
3. Existing Voice AI agents continue to function as before, with

   no changes required to the widget UI or embed code

   .

4. If an existing agent is switched to a supported S2S model, widgets already using that agent will update automatically.
5. The existing Chat Widget calling experience remains unchanged, including:

• Two-way audio
• Mute / unmute
• Call status
• End-call controls

6. The S2S SDK is loaded only when a visitor initiates a call with an S2S agent, ensuring regular widget loads and existing Voice AI calls remain unaffected.

1. Enable

   Speech-to-Speech Voice AI

   from Labs.

2. Create a new S2S agent or switch an existing Voice AI agent to a supported S2S model.

3. Navigate to

   Chat Widget → Agent

   .

4. Select the required Voice AI or S2S agent from the unified dropdown.

5. Save the widget.
6. The widget will automatically launch the correct voice experience when a visitor starts a call.

If a cloned agent is used instead, select the new cloned agent within the Chat Widget builder.