AI assistants

You can connect an AI assistant to Arvald and then just ask it things.

Who have I got booked in on Thursday, and has everyone paid?

Move Tuesday's 6pm squad to 7pm and tell me who that clashes with.

The assistant reads and writes through the same API the dashboard uses, as you. It sees exactly what you can see, for the brands you are a member of, and nothing else. There is no separate permission model to configure and nothing to grant beyond signing in.

What this is

Arvald runs a Model Context Protocol server, the standard way an AI assistant connects to an outside system. Claude and ChatGPT both speak it, so connecting is a matter of pasting one address and signing in.

The address is:

https://mcp.arvald.com

What to read

• Connect an assistant covers Claude and ChatGPT, step by step.
• What it can do is the full list, generated from the API itself.
• Modes and safety is the one to read before you let it change anything. It covers practising in your sandbox, what happens with refunds, and what disconnecting does and does not do.

Before you start

This is new, and worth being straight about:

• The assistant can do anything you can do, including issuing refunds and cancelling sessions. It is not a read-only view of your business.
• Assistants get things wrong. Check anything that moves money before you confirm it.
• Your booking and client data goes to whichever assistant you connect, which is Anthropic for Claude or OpenAI for ChatGPT. That is your choice to make, and their terms cover what they do with it.

If you want to try it without any of that mattering, connect it to your sandbox first. Every brand has one, it is full of fake data, and no money can move in it. That is modes and safety.