Google has opened a way for software to build notebooks in Gemini Notebook without a person clicking through the app. According to Google’s developer documentation, a program can create a notebook and fill it with sources: a web page by address, a YouTube video, plain text, or Markdown.
The quickstart is two calls. The first creates a notebook with a display name. The second attaches a source, and the example points at Google’s own Gemini API docs page. Requests go to notebook.googleapis.com, carry an access token, and name a Google Cloud project in a header. Samples come in Python, JavaScript, and curl.
Google also lists an MCP server, the connector standard that lets AI assistants call outside tools, next to the API. Both sit in an early access public preview, and only personal Google accounts work for now. Workspace accounts, which is what most companies run on, are not supported yet.
That limit decides who can try it. A team that wants to feed its research into notebooks automatically cannot do it from a work login today, so this is a prototype tool until Workspace support arrives. The page links to a roadmap but gives no date for it.
Published in Google’s developer documentation, last updated 2 November 2026.
