NotebookLM
NotebookLM Enterprise is Google's licensed, source-grounded notebook service for organizing sources in global-location deployments.
Verdict
Common use cases
- Turn meeting notes into audio briefings
- Synthesize research docs into executive summaries
- Convert customer feedback into insight reports
- Generate podcast-style overviews from sprint retrospectives
- Archive project knowledge in searchable notebooks
Integration
- Vendor
- NotebookLM
- Category
- other
- Auth
- OAUTH2
- Tools
- 11
- Composio slug
notebook_lm
Tools
- Add Text Sources
Add one or more named raw-text sources to a NotebookLM Enterprise notebook.
- Create Audio Overview
Start asynchronous generation of a NotebookLM Enterprise audio overview from selected notebook sources. Each notebook can have only one audio overview; if one already exists, Google returns it and ignores new generation options. The request
- Create Notebook
Create a NotebookLM Enterprise notebook in a licensed Google Cloud project and location.
- Delete Audio Overviewdestructive
Permanently delete an audio overview from a NotebookLM Enterprise notebook. If the audio overview is still being generated, this permanently cancels generation. This cannot be undone.
- Delete Notebooksdestructive
Permanently delete one or more NotebookLM Enterprise notebooks by notebook ID.
- Delete Sourcesdestructive
Permanently remove one or more sources from a NotebookLM Enterprise notebook.
- Get Notebook
Retrieve a NotebookLM Enterprise notebook and its current source metadata.
- Get Source
Retrieve a notebook source's ingestion status, metadata, and failure reason.
- List Recent Notebooks
List up to 500 NotebookLM Enterprise notebooks ordered by when the connected user last viewed them. Google currently omits continuation tokens, so accounts with more than 500 notebooks may not expose older results.
- Share Notebook
Grant, change, or remove account access to a NotebookLM Enterprise notebook. Notification emails are disabled by default; setting notify_via_email=true sends email and cannot be undone.
- Upload Source File
Upload a plain-text file as a source in a NotebookLM Enterprise notebook.
Setup
Setup guide
- 11. In Switchy, navigate to Settings → Integrations and click 'Connect' next to NotebookLM. 2. You'll authenticate via OAuth with your Google Workspace account — grant the requested scopes for notebook creation, source management, and audio generation. 3. Ensure your Google Cloud project has NotebookLM Enterprise API enabled and your account has the necessary IAM permissions (typically 'NotebookLM User' role). 4. After authorization, Switchy confirms the connection and displays your project ID. 5. Open any Project in Switchy and type '@NotebookLM create a notebook called Q1 Research' to test — the MCP should return a notebook ID confirming creation. 6. Add sources by @mentioning NotebookLM again with 'add this text as a source' followed by your content. 7. Generate an audio overview by asking '@NotebookLM create an audio overview from all sources in notebook [ID]' — processing takes several minutes.
What teammates see: by default, memories from NotebookLM are scoped to the Project (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Projects have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Create Research Notebook
@NotebookLM create a notebook called 'Competitor Analysis Q1' and add these three sources: [paste your text snippets or URLs here]
Generate Audio Overview
@NotebookLM create an audio overview from all sources in notebook [notebook_id]. Make it focus on actionable insights for the product team.
Add Meeting Transcripts
@NotebookLM add these meeting transcripts as sources to notebook [notebook_id]: [paste transcript text]
Check Source Status
@NotebookLM get the status of source [source_id] in notebook [notebook_id]. If it failed, show me the error.
Clean Up Old Notebooks
@NotebookLM delete notebooks [notebook_id_1] and [notebook_id_2]. Confirm they're gone.
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@NotebookLM create a notebook called 'Q4 Product Research' and add our three latest customer interview transcripts as text sources
I've created the notebook 'Q4 Product Research' in your NotebookLM Enterprise workspace. All three interview transcripts have been added as separate text sources and are now being indexed. The notebook is ready for queries once indexing completes — typically within 1-2 minutes depending on source length. You can now ask me to synthesize themes across these interviews or generate an audio overview.
This example shows the core workflow: creating a notebook and populating it with sources in one step. Requires NotebookLM Enterprise with OAuth2 authorization scoped to your Google Cloud project. The MCP handles notebook creation and source ingestion, but the actual AI synthesis happens when you follow up with analysis prompts.
@NotebookLM generate an audio overview from the sources in my 'Product Roadmap 2025' notebook, then give me the playback link
I've started generating the audio overview for 'Product Roadmap 2025'. NotebookLM is processing your sources into a conversational two-host discussion format. Generation typically takes 3-5 minutes. I'll retrieve the audio once it's ready. Note: each notebook supports only one audio overview at a time — if you already had one, Google returns the existing file rather than creating a duplicate.
Audio overviews are NotebookLM's signature feature: AI-generated podcast-style discussions of your sources. This is an asynchronous operation; the MCP initiates generation but you'll need to poll or wait for completion. Useful for making dense research material more digestible, but the two-host format may feel artificial for some content types.
@NotebookLM retrieve all sources from the 'Competitor Analysis' notebook and summarize which ones failed to ingest, then delete any failed sources
I checked the 'Competitor Analysis' notebook and found 8 sources total. Two failed during ingestion: 'techcrunch-article.txt' (unsupported encoding) and 'competitor-deck.pdf' (file size exceeded 50MB limit). I've deleted both failed sources from the notebook. The remaining 6 sources are fully indexed and available for queries. You may want to re-upload the TechCrunch article with UTF-8 encoding.
This demonstrates read-then-write orchestration: the MCP fetches source metadata, the AI reasons about failure states, then the MCP cleans up. Useful for maintaining notebook hygiene in automated workflows. Be cautious with deletion — it's permanent and cannot be undone. Source ingestion limits (file size, format support) are enforced by NotebookLM, not the MCP.
Use-case deep-dives
When NotebookLM beats wiki sprawl for new hires
A 6-person startup hires its first remote engineer. The onboarding checklist points to 14 Notion pages, 3 Google Docs, and a Slack channel archive. The founder uses NotebookLM to create a notebook, adds all those text sources in one batch, and generates an audio overview the new hire can listen to during their commute. This works because the source count is manageable and the content is mostly stable. If your onboarding materials change weekly or live in tools NotebookLM can't ingest (like Confluence or Linear), you'll spend more time re-syncing sources than the audio saves. For teams under 10 people with Google Workspace docs, this is the fastest way to turn scattered context into a listenable brief.
NotebookLM for synthesizing interview transcripts at sprint scale
A product team runs 8 customer interviews in a two-week sprint. Each transcript is 4,000 words. The PM dumps all eight as text sources into a NotebookLM notebook and asks the AI to summarize themes. This scenario is ideal because the sources are finite, the content is unstructured narrative, and the team needs a single synthesis artifact, not ongoing Q&A. The audio overview feature is less useful here unless the PM wants to brief stakeholders asynchronously. If your research volume exceeds 20 interviews per cycle or you need to query specific quotes later, a dedicated research repository with tagging beats NotebookLM's flat-source model. For sub-10 interview sprints, this is the lowest-friction synthesis tool in the Google stack.
When NotebookLM handles one-time narrative assembly
A founder prepares for a board meeting. She has Q3 metrics in a spreadsheet, a strategy memo in Docs, customer feedback in email threads, and product updates in Slack. She creates a NotebookLM notebook, adds the text-based sources, and uses the AI to draft a narrative connecting the dots. This works because the task is episodic (once a quarter), the sources are heterogeneous, and the output is a one-time deliverable. The audio overview can brief co-founders who didn't write the deck. If your board prep involves live data dashboards or requires version control across multiple drafts, NotebookLM's lack of integration with BI tools and its single-audio-overview limit become friction points. For founders who write in Google Docs and need a quarterly synthesis layer, this is the right call.
Frequently asked
What can the NotebookLM MCP do in Switchy?
It lets your team create and manage NotebookLM Enterprise notebooks programmatically — add text sources, generate audio overviews, retrieve notebook metadata, and delete sources or entire notebooks. You can automate research workflows that normally require clicking through NotebookLM's web interface, like bulk-adding documents or triggering audio summaries from your Switchy workspace.
Do I need a Google Cloud project to use this MCP?
Yes. NotebookLM Enterprise runs inside a licensed Google Cloud project, so you need OAuth2 credentials tied to that project. The MCP uses standard Google OAuth; you'll authenticate once in Switchy, and it handles token refresh. If your team doesn't have NotebookLM Enterprise set up in GCP, this integration won't work.
Can it upload PDFs or web pages as sources?
No. The MCP only supports adding raw text sources via the Add Text Sources tool. If you need to ingest PDFs, Google Docs, or URLs, you'll have to extract the text yourself first or use NotebookLM's web interface. The API doesn't expose file-upload endpoints yet.
Why use this instead of just opening NotebookLM in a browser?
Automation. If your team generates dozens of research notebooks from the same source types — meeting transcripts, support tickets, internal docs — the MCP lets you script that in Switchy instead of manually creating notebooks and pasting text. You also get programmatic access to audio overview status, which the web UI doesn't surface well.
Who on the team should connect this integration?
Whoever owns your Google Cloud project with NotebookLM Enterprise enabled. That person needs OAuth credentials scoped to the project. Other teammates can trigger the MCP's tools in Switchy once it's connected, but the initial auth must come from someone with GCP admin or editor rights on the NotebookLM resources.