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

# Go Live with Autopilot

> Deploy your Cendra AI agent to handle live guest conversations. Choose between manual, semi-automated, or full autopilot mode based on your confidence level.

Once you've tested your AI agent in the Sandbox, you're ready to go live. Cendra supports three automation levels so you can start with full oversight and increase automation as you build confidence.

<Info>**Quick link:** Open Agent Settings in Cendra — [https://app.cendra.ai/agents](https://app.cendra.ai/agents) — select your agent, then click the **Settings** tab to configure your automation level.</Info>

## Automation Levels

| Level              | What happens                                             | Best for                                                 |
| ------------------ | -------------------------------------------------------- | -------------------------------------------------------- |
| **Manual**         | AI suggests responses, you review and send               | First week — learning how the AI handles your properties |
| **Semi-automated** | AI sends routine messages, flags complex ones for review | Building confidence — AI handles the easy stuff          |
| **Autopilot**      | AI handles everything, escalates edge cases to your team | Full automation — AI manages 80%+ of guest communication |

## How to Go Live

1. Navigate to **Agent Hub → Agents → your agent → Settings**
2. Select your **working mode**:
   * **24/7 Autopilot** — AI responds at all hours
   * **Business Hours** — AI responds during your configured hours
   * **Custom Schedule** — set specific days and times
   * **Semi-Auto** — AI drafts responses for your approval
3. Configure **response timing**:
   * **Response delay** — wait 0-10 minutes before responding (feels more natural)
   * **Cool-off period** — minimum gap between consecutive messages
4. Set **reservation status filters** — choose which booking stages trigger AI responses (inquiry, confirmed, arriving, hosting, etc.)
5. Optionally assign the agent to **specific properties** or leave it active for all

## Recommended Go-Live Strategy

1. **Week 1: Semi-Auto** — AI drafts every response, you approve before sending. Learn what the AI does well and where it needs help.
2. **Week 2: Autopilot with guardrails** — Enable autopilot but keep escalation rules tight. AI handles routine questions, escalates anything complex.
3. **Week 3+: Full Autopilot** — Loosen escalation rules as confidence grows. AI handles 80%+ of messages automatically.

## Monitoring

After going live, monitor AI performance from the [Dashboard](/dashboard/overview):

* **Automation rate** — percentage of messages handled without human intervention
* **Escalation rate** — how often the AI hands off to your team
* **Response time** — how quickly guests receive replies
* **Active issues** — messages that need attention

<Card title="Explore the Dashboard" icon="chart-line" href="/dashboard/overview">
  Monitor your AI agent's performance and identify areas for improvement.
</Card>