Copilot Notebooks API: Access, Limits and Integration Options

Copilot Notebooks in Microsoft 365 are built on top of OneNote, and OneNote has a well-documented API through Microsoft Graph. That means the content of a Copilot-generated notebook, its pages, sections and text, can be read and written programmatically through the same Graph endpoints used for any OneNote notebook. The Copilot AI features that generate notebook content, however, are not separately callable through a public API.

For keeping selected references and work decisions outside the notebook, tools like fenbs store them where any connected AI app can read them, independent of your Microsoft 365 access.

What Copilot Notebooks are

Copilot Notebooks is a feature within Microsoft 365 Copilot that lets you create structured, notebook-style documents with AI assistance. Copilot can generate the initial content from a prompt, organize it into sections and help you expand on any part. The resulting document is stored and shared through your Microsoft 365 account.

For how Copilot Notebooks compare with Gemini Notebook on features and fit, see copilot notebooks vs notebooklm. Neighboring guides cover how they relate to OneNote, to Copilot Pages and to Copilot agents.

The Microsoft Graph API and OneNote access

Microsoft Graph is the unified API surface for Microsoft 365 services, including OneNote. It is documented at Microsoft's developer portal and covers notebooks, sections and pages. With appropriate permissions, an application can list notebooks, create pages, read page content as HTML and add to existing sections.

Because Copilot Notebooks are stored as OneNote notebooks, their content is reachable through those same Graph endpoints. A page created by Copilot looks, to the API, like any other OneNote page. You can read its HTML, extract the text and pass it to another system.

Authentication uses standard Microsoft identity, OAuth 2.0 with Azure AD. The permissions you need, such as Notes.Read or Notes.ReadWrite, are standard delegated or application permissions. Check Microsoft's Graph documentation for the current permission list, as it can change between API versions.

What the Copilot AI layer does not expose

The Copilot AI features themselves, asking Copilot to generate a notebook from a topic, to summarize a section or to expand on a point, are not available through a documented public API you can call directly. They are features within the Microsoft 365 Copilot product, and Microsoft has not published endpoints that let external code trigger Copilot AI generation inside a notebook.

Copilot Studio, which does have its own documented API surface, is a different product focused on building AI-powered agents and workflows. See copilot studio api for that integration path. They share the Copilot name but serve different purposes.

This distinction matters if your integration goal is to programmatically trigger Copilot to write notebook content. Through Graph you can write content yourself to a notebook; you cannot ask Copilot to do the writing for you via API call.

What you can access through Graph today

Through the Microsoft Graph OneNote API you can:

• List all notebooks in a user's account or a SharePoint site.
• Create a new notebook or a new section within an existing one.
• Create or update pages with HTML content.
• Read page content as HTML, including content that Copilot generated.
• Search notebook content using the Graph search endpoint.

These capabilities are enough to build a read integration, such as pulling notebook content into another system, or a write integration, such as appending structured notes to a specific section. Microsoft's developer documentation covers the exact endpoints, request formats and permission requirements.

How Copilot Notebooks compare with other AI notebook tools on API access

If your goal is AI-assisted notebook content that also has an API, it is worth checking what each tool exposes. Google's Gemini Notebook (formerly NotebookLM) does not have a public developer API. Copilot Notebooks, via Graph, gives read and write access to content even though the AI generation layer is not callable. For a full comparison of Copilot Notebooks with ChatGPT Projects, see copilot notebooks vs chatgpt projects. For a Copilot-versus-ChatGPT overview, see copilot vs chatgpt.

The GitHub Copilot cloud agent, which is a different product from Microsoft 365 Copilot, has its own integration surface covered in github copilot app. Gemini Notebook's current feature set is in gemini notebook.

Keeping references and decisions outside the notebook

A notebook is useful for organizing reference material and AI-generated drafts. It is less suited to holding active work decisions that need to reach multiple people and AI tools. A decision recorded in a notebook section may not reach a colleague using Claude or a developer using GitHub Copilot unless they search for it explicitly.

Keeping key decisions and where the work stands in a shared record outside the notebook means any connected AI can read them at the start of a session, regardless of which tool that person uses.

How to do this with fenbs

1. Create a free account and add the project your notebook belongs to.
2. Copy the key decisions and constraints from the notebook into Decisions and rules in fenbs. Add the reason behind each one.
3. Record the current task and its state in Where we left off.
4. Add links to the relevant notebook sections under Links and listings.
5. Connect a supported AI assistant from Integrations and ask it to read the project context before working, so it starts with the agreed decisions.

fenbs does not connect to Microsoft 365 directly and does not import notebook content. It holds the decisions and references you choose to save, keeping them available to any connected AI tool regardless of your notebook access.