Connect your AI model
How to connect an AI assistant such as Claude to your Coassemble workspace over MCP, and what it can do once connected.
Yes. You connect an AI assistant such as Claude to Coassemble over MCP (the Model Context Protocol). Create a connection in your Settings, copy the connection URL into your assistant, and sign in to Coassemble once to authorise it. From then on the assistant can create and manage courses directly in your workspace, using over 50 tools covering course creation, editing, publishing and analytics.
How do I connect an assistant?
- Open Settings in Coassemble and go to the MCP section.
- Create a connection and copy the connection URL.
- In your assistant, add that URL as a remote MCP server. Where you do this depends on the assistant.
- Sign in to Coassemble when prompted, to authorise the connection.
- The assistant appears in your list of authorised connections, and you're ready to go.
It's one sign-in. You don't need an API key, and you don't need to paste credentials into the assistant.
What can a connected assistant do?
Close to what you'd do yourself in the app:
- Create courses from a brief or from a document.
- Edit content screen by screen and block by block, including individual quiz questions.
- Restructure courses, reorder screens, and organise courses into collections and folders.
- Publish and share, including getting a share link.
- Read analytics, including course and screen level results.
- Work with branding, themes, translations and narration.
- Place and generate images, including images already in your workspace library.
Which assistants can I connect?
Any assistant that supports connecting to a remote MCP server and signing in. Because MCP is an open standard rather than a private integration, this isn't a fixed list, and support depends on the assistant rather than on Coassemble.
What is MCP?
MCP, the Model Context Protocol, is an open standard for connecting AI assistants and agents to external tools and data, so an AI tool can work with Coassemble through a documented, permissioned connection rather than a one-off custom integration.
Driving the agent from your own product
If you deliver Coassemble inside your own product or platform, your application can drive the same agent rather than a chat assistant doing it. The caller is identified using the same client and user identifiers your existing integration already uses, rather than a Coassemble login.
You get three controls over what it can do, which matter when you're exposing it to your own users:
- Course scoping. Limit the agent to a single course, so it can't reach anything else in the workspace.
- Tool selection. Choose which of the tools are available to it, rather than exposing all of them.
- Read-only mode. Let it read and answer questions without being able to change anything.
Frequently asked questions
Does this change which AI builds my courses?
No. Your assistant drives Coassemble and calls the right tools, but course generation runs on Coassemble's own AI, so a generated course is the same whether you asked in the app or through your assistant.
How do I see or remove a connection?
Your Settings list every assistant you've authorised, each with its name and icon. Disconnect any of them at any time and it immediately loses access to your workspace. The name and icon come from the application itself when it connects, so treat them as a label rather than proof of identity. If you see a connection you don't recognise, disconnect it.
Does an assistant use my AI allowance?
Yes. Anything that consumes your workspace allowance still consumes it when an assistant does it. Before spending allowance the assistant asks you to confirm and tells you what it will cost, so nothing is spent without your say-so.