Prerequisites
- A published agent.
- A widget deployment whose status is active.
- A website where you can add the embed code and test microphone permission.
Add the widget
- Open Deployments, create a widget deployment for your agent, and activate it.
- On its detail page, copy the Script Embed or Iframe Embed code. Copying from the page preserves the actual deployment ID and current engagement options.
- Paste it into a staging page, load that page over HTTPS, start a call through a user gesture, and confirm that the visitor can speak and hear a reply.
/embed/widget/YOUR_DEPLOYMENT_ID and must allow microphone; autoplay. When require user interaction is enabled, use the iframe’s generated companion widget_iframe_bridge.js script too; older iframe snippets without it cannot pass the host-page click to the iframe.
Engagement settings
The deployment can offer an incoming website call after an auto-call delay, require a prior page interaction, play an offer sound, and respect a visitor’s decline preferences. The delay starts when the widget loads; if Require user interaction first is enabled, it starts after the first click, key press, or tap. Interaction does not skip the delay. The offer itself is local to the browser; a voice session starts only when the visitor accepts and grants microphone access. For iframe hosts,?auto_call_delay=30 overrides the delay; for script embeds, use data-auto-call-delay="30". 0 disables auto-call for that host. Configure retry preferences in the deployment rather than implementing repeated offers in your host page.