Give Claude Code Full NotebookLM Access

One connection gives Claude Code full programmatic access to NotebookLM, including features the web UI doesn't have. Claude commands the research, Google's servers do all the heavy lifting, and your Claude token balance stays untouched. Setup takes 3 commands.

What You'll Learn

• 3-command setup (pip install plus login plus skill install)
• Zero token cost: Google handles all research processing
• Full CLI access including features beyond the web UI
• Prompt templates for every output type plus 3 real-world use cases

Why This Connection Matters

NotebookLM is the best research tool Google has ever built. It finds the highest-quality sources on any topic, reads all of them, and creates summaries you can actually use. But it lives in a browser tab. You have to switch apps, copy-paste, and manually manage everything.

Claude Code is the best AI assistant for getting real work done. It builds, automates, and executes. But when it needs to research something, it's limited to web searches and its own training data, and every query costs you tokens.

Connect them with notebooklm-py and you get something neither can do alone: Claude Code commands NotebookLM to do deep research, then generates audio overviews, infographics, editable slide decks, mind maps, flashcards, quizzes, and more, all from one prompt, all without leaving the terminal. Google's servers do all the heavy lifting. Zero Claude tokens spent on processing.

Setup (3 Commands)

Step 1: Install notebooklm-py. Open your terminal and install the package. This gives you a full command-line interface for NotebookLM with every feature, including things the web UI doesn't expose.

pip install "notebooklm-py[browser]"
playwright install chromium

Step 2: Log in to your Google account. Run the login command. It opens a browser window. Sign in with your Google account and you're authenticated.

notebooklm login

Step 3: Install the Claude Code skill. This installs a skill that teaches Claude Code how to use every single NotebookLM command. After this, Claude knows how to create notebooks, add sources, generate content, download files, and more, all automatically.

notebooklm skill install

Verify the connection. Open Claude Code and ask it to list your notebooks. If it returns your NotebookLM dashboard, you're connected.

Troubleshooting: if Claude Code can't connect, run notebooklm login again. The authentication cookies expire periodically, so you'll need to re-login occasionally. You can also run notebooklm auth check (with the test flag from the docs) to diagnose any issues.

What You Can Create

Once connected, Claude Code has full programmatic access to NotebookLM, including features the web UI doesn't expose. Each notebook supports up to 50 sources (more on paid plans). Here's every output type.

Audio overviews (4 formats). NotebookLM generates podcast-style conversations between two AI hosts. Four formats: deep-dive (thorough analysis), brief (quick summary), critique (balanced evaluation), and debate (opposing viewpoints). Three length options. 50+ languages. Takes about 3 to 5 minutes to generate. Perfect for learning on the go.

Research [YOUR TOPIC] using NotebookLM.
Create a notebook, add the best sources,
and generate an audio overview in
debate format. Download as MP3.

Infographics (11 styles). Visual summaries in a single shareable image. 11 styles: bento grid, professional, editorial, instructional, sketch-note, scientific, bricks, clay, anime, kawaii, and auto. Three orientations (landscape, portrait, square) and three detail levels. Takes about 2 minutes.

Create a notebook on [YOUR TOPIC],
add the best sources, and generate
an infographic using the bento-grid
style in portrait orientation.

Slide decks (editable PowerPoint). Full presentation decks generated from your research. Download as PDF or editable PPTX. The web UI only offers PDF, but notebooklm-py gives you the editable PowerPoint file. You can even revise individual slides with natural language prompts after generation.

Video overviews (9 visual styles). Video summaries with visual elements and narration. Nine visual styles including whiteboard, anime, kawaii, watercolor, retro-print, heritage, and paper-craft. Three formats: explainer, brief, and cinematic.

Mind maps, quizzes, flashcards, and more. Mind maps export as JSON for visualization tools. Quizzes and flashcards export as JSON, Markdown, or HTML. The web UI only shows an interactive view, but notebooklm-py lets you download them in structured formats. Data tables export as CSV. Reports generate as briefing docs, study guides, or blog posts.

YouTube as a source. Paste any YouTube URL as a source and NotebookLM pulls the full transcript automatically. Research a topic by adding the top 10 YouTube videos on it. NotebookLM reads every transcript and synthesizes the key insights across all of them.

3 Real-World Use Cases

1. Weekly competitive intelligence. Every Monday, tell Claude to research what your top 3 competitors shipped last week. It builds a NotebookLM notebook with their blog posts, press releases, product updates, and social media mentions. Then it generates an audio overview you can listen to during your commute and an infographic you can share with your team in Slack. 15 minutes of research time replaced by one prompt.

2. Client-ready research packages. Running an agency or consulting business? Before every client call, tell Claude to research the client's industry trends, generate an editable slide deck with the key findings, and create an infographic summarizing the data. Download the PPTX, add your branding, and present. Show up to every meeting with custom research that looks like your team spent days on it.

3. Content research on autopilot. Creating content about a topic you're not an expert in? Tell Claude to research it via NotebookLM. Add the top YouTube videos as sources so it reads the transcripts, then generate a debate-format audio overview to hear both sides. Ask follow-up questions until you understand the topic. Then have Claude write your script, blog post, or newsletter, grounded in sources it actually read.

Prompt Templates

Copy-paste these into Claude Code. Replace [YOUR TOPIC] with whatever you're researching.

Full research package

Research [YOUR TOPIC] using NotebookLM.
Create a new notebook, find and add the
best sources (include YouTube videos),
then generate:
1. An audio overview (deep-dive format)
2. An infographic (bento-grid style)
3. A slide deck (download as PPTX)
Wait for each one to finish and
download the results.

YouTube deep-dive

Find the top 10 YouTube videos about
[YOUR TOPIC]. Create a NotebookLM
notebook and add each video URL as a
source. Generate a debate-format audio
overview that covers the different
perspectives across all 10 videos.

Competitor research

Research [COMPETITOR NAME] using
NotebookLM. Find their latest blog
posts, product updates, and news.
Create an infographic summarizing
what they shipped this month and
generate an audio overview.

Study package

I need to learn [YOUR TOPIC] fast.
Create a NotebookLM notebook with
the best sources. Generate:
1. An audio overview to listen to
2. Flashcards (download as markdown)
3. A quiz to test my understanding
4. A study guide report

Go Deeper

• Printable PDF version of this guide: https://drive.google.com/file/d/1GOy0Vg-wFnpUCjx_eAJeMHkjpWqve20k/view?usp=drivesdk

Want the whole system installed live, with Cooper on the call? Build Your Own AI, the two week sprint runs the first two weeks of August.

Or get a new working tool every Monday inside The AI Operator community on Whop.