Skip to main content

Connecting Cobot to Claude, ChatGPT or Other AI Assistants Using MCP

Manage your space in plain language by connecting Claude, ChatGPT or other AI assistants to Cobot.

Written by Laura

The Cobot MCP server lets you connect an AI assistant, such as Claude or ChatGPT, to your Cobot space using the Model Context Protocol (MCP). Once connected, you can ask the AI to handle everyday admin tasks in plain language – like checking desk availability or booking a room – all in the chat.

Message request to book the Hot Desk for Carolina from 3:30pm to 5:30pm tomorrow, entered in Claude's chat interface.

💡 Good to know: You stay in control the whole time. The assistant can only see the spaces you manage, only uses a limited set of permissions, and never receives your password.


Getting Started

First, make sure you have the following:

  1. A Cobot account (trial or paid subscription) where you are an admin

  2. An AI assistant that supports remote MCP servers, such as Claude or ChatGPT.

That is all you need. There is nothing to install inside Cobot.


Setting Up the Connection

You connect everything from inside your AI assistant, not from the Cobot admin portal. The exact steps vary slightly by assistant, but the flow is the same everywhere:

  1. In your AI assistant, open the settings area where you add connectors or MCP servers.

  2. Add the Cobot MCP server URL and paste in the Cobot server address: https://api.cobot.me/mcp

  3. Your assistant opens a Cobot login window. Log in and review the access it is requesting.

  4. Approve the request. Your AI assistant is now connected to your Cobot space.

Once connected, try asking your assistant something simple to confirm it is working, for example: "Which desks are free this afternoon?"

💡 Pro Tip: If you manage multiple space, your assistant can see all of them. So make sure to name the specific space in your request in order for the assistant to act on the right one.

Now jump straight to What Can you Ask your AI Assistant or review more detailed setup instructions for Claude and ChatGPT below.


Detailed Instructions for Claude

1. Log in to Claude.

2. Navigate to Customize » Connectors.

3. Click "+" then "Add custom connector".

Claude customization menu showing the Connectors section with a plus button to add a new connector.

4. Add the Cobot MCP server URL: https://api.cobot.me/mcp

Claude connector setup form with a field to enter the remote MCP server URL.

5. Finish configuring your connector by clicking "Add.”

6. After configuring it, click "Connect" and sign in with your Cobot credentials.

7. After connecting, you can specify which tools require permission and which the model can use directly.

MCP configuration screen for Cobot showing a list of available tools with toggles to enable or require permission approval for each tool.


Detailed Instructions for ChatGPT

  1. Log in to ChatGPT.

  2. Open Settings.

  3. Select "Security and login".

  4. Turn on Developer mode.

ChatGPT Settings page showing Security and login options with Developer mode toggle enabled.

Navigate to ChatGPT Plugins to add your MCP server.


5. Click the "+" button to add a new plugin.

ChatGPT Plugins interface with a plus button to add a new MCP server.


6. Select "Create a custom plugin" or "Add from URL" (exact wording depends on current ChatGPT UI).

7. Enter a user-facing name and description.

8. Paste the Cobot MCP server: https://api.cobot.me/mcp (including the /mcp path).

9. Create the connection.

10. Review the tools and metadata discovered from the server.


What Can you Ask your AI Assistant?

Once connected, you can ask your AI to help with bookings, resources, members, and invoices.

Message request to move the start of Carolin's  Hot Desk booking from 8am to 6am, entered in Claude's chat interface.

Some examples:

  • "Which desks are free this afternoon?"

  • "List today's bookings"

  • "Book the small meeting room for Jane tomorrow 10–12"

  • "Move today's 2pm boardroom booking to 3pm."

  • "Sarah just left — check her out."

  • "What's booked in Room A next week?"

  • "Any visitor bookings today?"

  • "Which resources are allocated long-term, and to whom?"

Behind the scenes, the assistant uses a small set of tools to do this:

  • See your account - which user is logged in and which spaces you administer

  • View resources - rooms, desks, and other bookable resources in a space

  • Check availability - which resources are free in a given time range

  • View members - the members of a space

  • Create bookings - book a resource, optionally on behalf of a member

  • Review bookings - look up a single booking or list bookings over a date range

  • Review invoices - list and search invoices

💡Pro Tip: The AI assistant works best when your request is specific. Including a space name, a date, and a time range gives you the most accurate results.


Security and Access

Your Cobot data stays protected throughout.

  • Login uses OAuth 2.0 with PKCE, the same secure method as the Cobot API. No passwords or API keys are ever shared with the assistant.

  • Access is limited to a small set of permissions, covering only bookings, resources, memberships, and invoices – and only for the spaces you manage.

  • You can revoke an assistant's access at any time under your authorized applications.

⚠️ Apply the same data protection and confidentiality standards you would to any admin session containing personal information


Next Steps

The connection is new, and the set of available tools will grow over time. If you have feedback or an idea for something you would like your assistant to handle, let us know.
For developers building their own MCP client, full technical details, including transport and OAuth discovery, are on the Cobot MCP server page.

Did this answer your question?