Your AI Agent can run in one of two modes: Supervised Mode or Autopilot.

In both, the AI handles the whole conversation, drafting replies and taking actions like escalating, closing, and applying labels. The difference is whether a human approves each action before it goes out.

Supervised Mode is a 30-day review period. The AI prepares every reply and action, but nothing is sent or executed until a human approves it; replies go out under the human's name, and a human always stays the ticket owner. It's a defined window to watch the AI handle your real conversations before you commit to full automation. When the 30 days are up, the AI simply stops drafting. There's no charge and no automatic upgrade, and your configuration is saved: you either switch on Autopilot to keep going (or contact your CSM for more time).

Autopilot is full automation. The AI handles conversations on its own, replying, escalating, closing, and labelling, without waiting for approval.

Which should I choose?

Start with Supervised Mode to see how the AI handles your real conversations during the 30-day review period, then switch to Autopilot once you're confident. You can change the mode per AI Agent at any time. (Neither mode is available for Voice agents or AI Agents embedded in an AI Journey.