> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trellistech.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Autopilot

> Review the AI’s replies, improve missing knowledge and understand how requests reach your team when they need human attention.

[Resources](/resources/index) / AI & knowledge · 6:20 · English

<div data-trellis-resource-video="autopilot">
  <video controls playsInline preload="none" poster="/images/support-resources/autopilot.webp" aria-label="Autopilot" style={{ width: '100%', borderRadius: '12px' }}>
    <source src="https://app.trellistech.com/v/autopilot" type="video/mp4" />
  </video>
</div>

## Video recap

Review the AI’s replies, improve missing knowledge and understand how requests reach your team when they need human attention.

### Start with suggested replies

The first example uses a workspace where automatic replies are off. Review the proposed response and open the AI activity to inspect the information and actions behind it before deciding what to send.

### Teach the AI and review the result

Use Teach AI to explain what was wrong or provide the missing property-specific information. The update runs in the background and may ask for clarification. Check the resulting knowledge changes rather than stopping after submitting feedback.

### Handle escalations

The second example shows a request the agent cannot resolve on its own. It acknowledges the guest’s request and creates an escalation; configured workflows notify the relevant team through their chosen channels.

### Follow the sources

When a reply includes a source reference, open it to review the information used. The video encourages regularly reviewing replies and correcting the underlying knowledge as part of the team’s workflow.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.