> ## 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.

# Sandbox — Test Your AI

> Cendra's Sandbox is a testing environment where you can send messages to your AI agent with mock reservation data before deploying to live guests. Same pipeline as production.

Cendra's Sandbox is a testing environment where you can send messages to your AI agent with mock reservation data before deploying to live guests. The Sandbox uses the exact same AI pipeline as production — what you see in testing is what guests will experience.

<Info>**Quick link:** Open Sandbox in Cendra — [https://app.cendra.ai/sandbox](https://app.cendra.ai/sandbox)</Info>

## Why Test in the Sandbox

* **Verify accuracy** — confirm the AI answers common questions correctly using your property data
* **Test guardrails** — make sure AI Rules prevent unwanted responses
* **Check escalations** — verify that escalation rules trigger when they should
* **Tune tone** — confirm the AI's communication style matches your brand
* **Build confidence** — feel comfortable before switching to autopilot

## Mock Reservation Setup

Configure a test reservation to simulate different guest scenarios:

| Setting                 | Options                                                                                                         |
| ----------------------- | --------------------------------------------------------------------------------------------------------------- |
| **Property**            | Select from your synced properties                                                                              |
| **Check-in date/time**  | Any date — test pre-arrival, during-stay, and post-stay scenarios                                               |
| **Check-out date/time** | Any date                                                                                                        |
| **Reservation status**  | Inquiry, Follow-up, Confirmed, Arriving, Check-in today, Currently hosting, Check-out today, Post-stay, Expired |
| **Booking channel**     | Airbnb, Booking.com, Direct, Other                                                                              |
| **Guest count**         | Adults and children                                                                                             |
| **Payment status**      | Paid or unpaid (affects payment reminder behavior)                                                              |
| **Message date/time**   | When the test message was "sent" (affects time-sensitive responses)                                             |

## How to Test

1. Navigate to **Agent Hub → Test Agent**
2. Select a property and configure the mock reservation
3. Type a message in the chat input
4. Review the AI's response
5. Adjust knowledge, rules, or agent settings as needed
6. Repeat until you're satisfied

<Frame caption="Sandbox — test your AI agent with mock reservation data and see responses in real time">
  <img src="https://mintcdn.com/botelaiinc/8mqmF829G5YGubLh/images/10-sandbox.png?fit=max&auto=format&n=8mqmF829G5YGubLh&q=85&s=5edd541a640af6c9d5a381efe6ff7309" width="3396" height="1774" data-path="images/10-sandbox.png" />
</Frame>

## Recommended Test Scenarios

| Test                                 | What to verify                                     |
| ------------------------------------ | -------------------------------------------------- |
| "What time is check-in?"             | Correct property-specific check-in time            |
| "What's the WiFi password?"          | Retrieves from knowledge base                      |
| "Can I bring my dog?"                | References pet policy                              |
| "I want a refund"                    | Escalation triggers (if configured)                |
| "Where's the nearest grocery store?" | Local recommendations from knowledge base          |
| Message in Spanish/French            | Language matching works correctly                  |
| "The shower is broken"               | Maintenance escalation triggers                    |
| "Can I check in at 10am?"            | Early check-in upsell or escalation per your rules |

## Sandbox Limitations

* Messages are not sent to real guests
* No actual PMS data is modified
* Upsell offers are simulated but not charged
* The Sandbox does not connect to WhatsApp or email — it's a standalone chat interface