More to configure

The collapsed section at the bottom of the agent page holds seven panels. What each one is for, and the order worth opening them in.

The panel behind how your agent thinks and sounds: provider, model, response style, transcription and voice. What each control does and which one to reach for.

What happens when an answering machine picks up, how long a call can run, when the agent hangs up, where a call goes when the agent cannot finish it, and the two sliders that set its conversational timing.

Sentiment, summaries, natural speech and the follow-ups that fire after a call ends. Six switches, what each costs you, and which are worth leaving on.

The Tools panel is where an agent gets the ability to do things: attach integrations, set custom values it can use mid-call, and connect tracing for debugging.

Turn conversations into rows. Which fields to collect, what Confirm and Required do, and the one thing about keys that will bite you later if you get it wrong.

The node canvas for calls whose steps have to happen in order. Seven node types, five templates, two ways to generate one, and one control that decides whether any of it reaches a live call.

The Embed panel turns an agent into a voice widget you can drop onto a page, so visitors can talk to it without picking up a phone.