How-to guide
How to connect Google Drive to Claude
A step-by-step guide to connecting Google Drive to Claude with KBrain, so Claude can answer from your actual documents instead of files you paste into each chat.
Turn Google Drive into an AI knowledge base
To connect Google Drive to Claude, you point a knowledge layer at a Drive folder and expose it to Claude over MCP. Claude cannot read your Drive on its own. With KBrain in between, you connect a folder once, add the generated MCP endpoint to Claude, and Claude can then query those documents live during any conversation, with nothing to upload.
Why Claude cannot just open your Drive
Claude has no standing access to your Google Drive, and giving it raw access to every file would not help anyway. A folder of hundreds of documents is far larger than any context window, and dumping it all in would bury the relevant passage in noise. What Claude needs is retrieval: the ability to pull the few relevant paragraphs from the right document at the moment it answers. That is what a knowledge layer over MCP provides.
The setup, step by step
- Pick the Google Drive folder that holds the knowledge Claude should reference, a client folder, a policy set, research notes, onboarding material. A focused folder retrieves more precisely than your whole Drive.
- Connect that folder to KBrain and authorize Google Drive access. KBrain indexes the documents into a private, queryable brain and generates an MCP endpoint for it.
- Add the endpoint to Claude as an MCP connector, in Claude's connectors or MCP settings, and authorize it if prompted.
- Ask Claude a question the folder covers. Claude queries the brain, pulls the relevant passage, and answers from your document instead of its training data.
What changes once it is connected
- No more re-uploading. The knowledge persists across every conversation, not just the one where you attached a file.
- Always the current version. When a document in the folder changes, Claude sees the update on its next query, with nothing to re-sync by hand.
- One source, every assistant. The same KBrain endpoint works in ChatGPT and any MCP client, so you curate the folder once and use it everywhere.
Google Drive stays your system of record. KBrain does not move or copy your workflow out of Drive; it exposes a focused, private, retrievable view of a folder to Claude.
Common issues
- Claude answers generically. Confirm the connector is enabled for that conversation and that the question falls within what the connected folder covers.
- A new document is not showing up. Give indexing time to pick up recent changes, and check that the file lives inside the connected folder rather than a sibling one.
- Access was denied. Private brains are scoped to whoever created them. To let a teammate query the same Drive brain, grant them access rather than sharing your own connection.
Turn Google Drive into an AI knowledge base
Subscribe to KBrain, connect a Google Drive folder, and give Claude, ChatGPT, or any MCP compatible assistant access to curated, private knowledge.
Frequently asked questions
Can Claude read my Google Drive directly?
Not on its own. Claude reaches external data through MCP connectors. KBrain connects a Drive folder and exposes it as an MCP endpoint that Claude can query, so Claude answers from your documents without direct Drive access.
Do I have to connect my whole Google Drive?
No, and you should not. Connect a focused folder scoped to a clear purpose. Retrieval is more precise over a curated folder than over an entire Drive, and it keeps unrelated files out of scope.
Does connecting Drive to Claude move my files?
No. Google Drive stays the system of record. KBrain indexes the selected folder into a private brain for retrieval and reads updates on the next query. Your files stay where they are.