Overview: NotebookLM Course
Follow this NotebookLM course through the Gemini Notebook workflows, with source-based research, synthesis and advanced learning exercises.
Google Gemini Notebook (NotebookLM) – Advanced Guide and Masterclass
TL;DR
- This is a full, self-contained course on Gemini Notebook, the tool Google renamed from NotebookLM on July 16, 2026. It’s the same tool and the same free plan under a new name, and you’ll find it at notebook.google.com.
- We’re not sending you anywhere else to “finish learning it.” Every module below, from your first notebook to Deep Research reports, Audio Overviews and multi-notebook workflows, is taught here, in full.
- It’s free to start: 100 notebooks and 50 sources each, with no credit card. One thing has changed since this course first went up: since September 2, usage limits are compute-based and refresh every five hours up to a weekly cap, replacing the old fixed daily counts.
- By the end of this course, you’ll be able to build a source-grounded research assistant, run Deep Research investigations, generate podcast-style Audio Overviews and automate recurring research, all without writing a single line of code.

Reviewed by Imran Khan Pathan, Editor at AI Tech Safar. I built out real notebooks for this course – dumped in PDFs, YouTube links and web sources, ran actual Deep Research sessions and generated Audio and Video Overviews – before writing a single instruction below. Every workflow here is something I ran myself, not a rewording of Google’s help docs. For the October 7 update, every limit, plan detail and feature claim was re-checked against Google’s own Gemini Notebook help center, its Workspace Updates blog and its announcement posts. Where a feature depends on your plan, I’ve said so plainly.
Last updated: October 7, 2026
What changed since the September version
- Usage limits: on September 2, Google replaced fixed daily counts (50 chats, 3 Audio Overviews and so on) with compute-based limits that refresh every five hours up to a weekly cap.
- Plans: Ultra splits into a $99.99 (20 TB) and a $199.99 (30 TB) plan, and Plus now costs $4.99.
- New: a secure cloud computer inside each notebook, notebooks inside the Gemini app, and September 15’s real-time conversations, mobile audio recording and interactive overviews.
- Corrections: Deep Research lives in the sources panel, not Studio; slide exports are PPTX or PDF; the model is Gemini 3.5-class, not plain Gemini 3; and the “same model on every tier” claim has been softened.
Who This Course Is For
This is written for people who already know Gemini Notebook exists but have never gone past “upload a PDF, ask it a question.” If that’s as far as you’ve gotten, you’re using maybe 20% of what the tool can do.
By the time you finish, you’ll know how to:
- Structure notebooks so the AI stops giving you vague, generic answers
- Run multi-source research investigations with Deep Research
- Turn a pile of documents into a podcast, a slide deck, a study guide and a quiz, all from the same source material
- Automate recurring research so your notebook keeps filling itself
- Avoid the mistakes that make most people’s notebooks hallucinate or give shallow answers
No prior AI experience needed, and no other tutorial required. Let’s start from the foundation.
Module 1: What Gemini Notebook Actually Is (The Part Everyone Skips)
Before you touch the interface, take in one idea, because it changes how you should use the tool: Gemini Notebook is not a general chatbot. It’s a source-grounded research assistant.
ChatGPT, Gemini’s regular chat and Claude all answer from their training data plus whatever you type. Gemini Notebook works differently. It answers only from the documents, videos and links you put into that notebook. If you didn’t upload it, the notebook doesn’t know it exists, and it tells you so instead of guessing.
That design is why Google built it in the first place. It started in 2023 as Project Tailwind, a prototype for research, studying and analysis, where “confidently wrong” is worse than “I don’t know.” It now has more than 30 million users and over 600,000 organizations. Google doesn’t publish which model version runs it, but reporting shows it moved to Gemini 3 in December 2025 and to Gemini 3.5 from June 2026, with the newest models reaching Ultra and Workspace customers first.
What that means for you: the quality of what comes out depends almost entirely on the quality and structure of what goes in. This course spends real time on that, because most tutorials skip it.
Module 2: Choosing Your Plan (Free Is Enough to Start)
You can’t buy Gemini Notebook on its own. It comes bundled with Google’s AI subscriptions, which Google reorganized at I/O on May 19, 2026. Here’s what each tier gives you:
| Plan | Price | Notebooks | Sources/Notebook | Usage allowance |
|---|---|---|---|---|
| Free | $0 | 100 | 50 | Standard |
| Plus (Google AI Plus) | $4.99/mo | 200 | 100 | 2x standard |
| Pro (Google AI Pro) | $19.99/mo | 500 | 300 | 4x standard |
| Ultra (Google AI Ultra, 20 TB plan) | $99.99/mo | 500 | 500 | 5x or 20x AI Pro, depending on the Ultra plan |
| Ultra (Google AI Ultra, 30 TB plan) | $199.99/mo | 500 | 600 | 5x or 20x AI Pro, depending on the Ultra plan |
What’s the same on every tier: every core Studio feature, including Deep Research, Audio Overviews, Video Overviews, Slide Decks, Reports, Flashcards and Quizzes, and a cap of 500,000 words or 200 MB per individual source. What paying buys is more room: more notebooks and sources, higher usage allowances, and earlier access to new features. Google says Pro and Ultra also get higher access to its Gemini models, though it doesn’t name the version. Cinematic Video Overviews, built with Google’s Veo 3 model, have their own plan rules, covered in Module 6.
Teams have other routes too. Fello AI reports that Gemini Notebook Plus is included in qualifying Google Workspace plans from Business Standard ($14 per user a month), and that Gemini Notebook Enterprise is sold through Google Cloud from roughly $9 per license with a 15-license minimum.
How Usage Limits Work Now
Until September 2, Google published fixed daily counts for each feature: on Free, 50 chats a day, 3 Audio Overviews, 3 Video Overviews and 10 Deep Research sessions a month. Plenty of guides still quote those numbers. Google’s own help center no longer does. Gemini Notebook now runs on a compute-based budget:
- Your allowance depends on how much work a request costs: prompt complexity, chat length, the number of sources, the model and the features you use.
- The allowance refreshes every five hours, until you reach a weekly limit.
- Open Settings → Usage to see what you’ve used. A note below the chat tells you when your limit resets, and the Studio panel shows the expected cost of an output before you generate it.
The practical effect: a quick chat question is cheap, and a long Video Overview over 50 sources is expensive. Plan your heavy jobs deliberately.
Course recommendation: start on Free. Everything in this course works there. You’ll only feel the ceiling if you run the tool professionally, every day, at volume.
Module 3: Building Your First Notebook Properly
Go to notebook.google.com and sign in with any Google account. Click New notebook. This is where most tutorials stop being useful: they tell you to “add a source” and say nothing about how to think about sources.
What You Can Add as a Source
- PDFs, Google Docs and Word documents – the most common source types
- Google Slides, Google Sheets and CSV files, plus images, which are read with OCR
- Web page URLs – the page content is pulled in directly
- YouTube video links – it imports the transcript, not the video, so lectures, interviews and long videos become searchable text with timestamped citations
- Audio files – upload a recording and it transcribes it and grounds itself in it
- EPUB files – full ebooks, supported for every user since March 2026
- Pasted text – copy and paste anything directly
The Rule Nobody Tells You: Curate, Don’t Dump
The instinct is to upload everything you have. Resist it. A notebook with 40 loosely related sources gives vaguer answers than one with 8 tightly relevant ones, because the model has to spread its attention across everything. Build one notebook per project or question, not one giant notebook for “everything I’m working on.” This is the biggest quality lever in the whole tool, and it costs nothing.
Use Discover to Find Sources You Didn’t Have
If you don’t have enough material yet, use Discover in the sources panel. Describe what you’re researching and it suggests up to about ten relevant web sources you can add with one click. That turns notebook-building from “gather everything manually” into a guided process.
Module 4: The Chat Panel – Prompting a Grounded AI Correctly
Because Gemini Notebook only answers from your sources, prompting it well is a different skill from prompting ChatGPT. Grounding doesn’t fix a lazy question; generic prompts still get generic answers.
Five Techniques That Change Output Quality
1. Check the citations before you trust anything. Every answer carries inline citations that point to the exact passage in your source. Click through and verify anything you’ll use in real work. It’s the tool’s main safety feature, and most people ignore it.
2. Chain follow-ups instead of restarting. Your chat history is saved between sessions and is private to you, even in a shared notebook. So you can leave a thread, come back a day later and keep building. Ask a broad question first, then narrow: “Now compare that to what source 3 says” or “Which of these claims has the weakest source support?”
3. Ask it to disagree with itself. Try: “Find any contradictions between my sources on this topic.” Because it’s grounded in several documents, it can surface where your sources conflict, which a single read-through would miss.
4. Interrogate the gaps. Ask: “What questions do my current sources leave unanswered?” That tells you exactly what to look for next and turns the chat into a research-gap finder, not just a Q&A box.
5. Set a goal for the notebook. Open the gear icon in the chat panel and write a role in the Goals box, up to 5,000 characters: a patient teacher, a rigorous PhD advisor, a skeptical editor. Every answer in that notebook follows it. You can also ask in chat for an Audio Overview, a Video Overview or a report, and it creates the output without you leaving the conversation.
Module 5: Deep Research – The Feature Nobody Teaches Properly
This is the most advanced capability in the tool and the one this course spends the most time on, because it’s underused.
What Deep Research Actually Does
Deep Research runs an investigation instead of a single-turn answer. It makes a research plan, searches the web, reads what it finds, and writes a citation-backed report while you keep working. Unlike a chat answer, you can inspect where its claims came from.
How to Run a Deep Research Session
- In your notebook, go to the sources panel and choose Web as the source type. (It isn’t in the Studio panel, which is where older guides, including earlier versions of this one, sent you.)
- Pick between Fast Research, which quickly surfaces sources for you to review, and Deep Research, which produces a full briefing.
- Write your question as a scope, not a fact lookup. Bad: “What is X?” Good: “Investigate the main arguments for and against X, who’s making each argument, and what evidence they cite.”
- Let it run in the background. It takes longer than a chat answer because it’s doing multi-step work.
- Review the report and its sources, then add them to the notebook with one click, so future chat questions can use what you just gathered.
Three Real Deep Research Use Cases
Competitive analysis: “Research how our three main competitors have priced this feature over the last 12 months, and what public criticism each has received.” Upload whatever internal documents you have first, then let Deep Research fill the external gaps.
Literature review: upload 10-15 papers on a topic, then ask: “Synthesize the current consensus and disagreements across these papers, and identify which findings the others cite most often.” That alone can save hours of cross-referencing.
Due diligence: before a big purchase, partnership or decision, try: “Research the track record, controversies and independent reviews of [company/product], separating verified claims from marketing language.”
Quota reality check: a Deep Research run is one of the more expensive things you can ask for. Spend it on questions that genuinely need multi-step research, not on something a normal chat question would answer just as well.
Module 6: The Studio Panel – Turning Sources Into Other Formats
This is where Gemini Notebook stops being “a chat with your documents” and becomes a content production tool. Everything in Studio pulls from the same grounded sources, so you never start over.
Audio Overviews (The Feature That Made This Tool Famous)
Click Audio Overview and it generates a podcast-style conversation between two AI hosts discussing your sources. Most people generate the default and stop. The better moves:
- Pick a format. Options include Deep Dive, Brief, Critique and Debate, and audio comes in more than 80 languages. You can download the result.
- Use the focus prompt. Before generating, steer it: “Focus on the financial risks in these documents” or “Explain this at a beginner level, avoid jargon.” A generic overview and a focused one are very different in usefulness.
- Use Interactive Mode. You can interrupt the hosts mid-conversation and ask a follow-up. They answer from your sources, then pick up where they left off.
- Generate more than one angle. Run it once for a general overview and again with a narrow prompt for a subtopic. Same material, two useful outputs.
Video Overviews
A narrated, slide-style video built from your sources, the visual sibling of the Audio Overview. You can choose Explainer or Brief, add a steering prompt, and pick a visual style such as Classic, Whiteboard, Watercolor, Retro Print, Heritage, Paper-craft, Kawaii or Anime.
Cinematic Video Overviews go further: fully animated, narrative videos built with Gemini, Nano Banana Pro and Veo 3. They’re English-only and for users 18 and over. Google’s March availability note lists Google AI Pro and Ultra plus eligible Business and Enterprise plans, while several trackers still describe them as Ultra-only, with daily caps of 2 on the 20 TB plan and more on the 30 TB plan. If Cinematic doesn’t appear in your Studio panel, it isn’t enabled on your account yet.
Mind Maps, Reports, Data Tables, Flashcards & Quizzes
These are built for learning and analysis:
- Mind Maps – a clickable visual of how the concepts in your sources relate. Useful for untangling a dense document before you read it linearly.
- Reports – structured documents such as study guides, briefing documents and blog-post drafts, from templates or your own prompt.
- Data Tables – turn qualitative text, such as competitor descriptions or study summaries, into a comparison table you can export to Google Sheets.
- Infographics – choose from ten styles, including Sketch Note, Scientific, Bento Grid and Clay, or let it pick.
- Flashcards and Quizzes – generated from your sources, with a “Got it / Missed it” sort, a shuffle, and a results screen that reruns the cards you missed. Progress saves across sessions.
Slide Decks: Revise, Then Export
Generate a slide deck from your sources, then revise it: give feedback on individual slides, such as “make slide 5 more concise,” and the slides regenerate as a new deck in Studio. Export as PPTX or PDF. One caution from testers: in PowerPoint the exported slides can behave like image layers, not fully editable objects, so heavy layout changes may need another tool.
Module 7: Power-User Workflows (Where This Course Goes Further Than Anyone Else)
Automate Recurring Research with Workspace Studio
On Workspace accounts, Google Workspace Studio has an Add a source to Gemini Notebook step. Build it into an automated flow and it adds text, Drive file links, YouTube links or web URLs to a notebook on its own, each time the flow runs. That turns a one-off notebook into a living research feed for something you track continuously, such as a competitor, a market or an ongoing project. There’s also an Ask NotebookLM step, which lets an automation draw on a notebook for grounded answers.
Copy a Notebook Instead of Rebuilding It
You can copy an entire notebook, including its sources and Studio items, if you have copy permission for it. Use this to build a clean template notebook once, then duplicate it for every new project.
Share Research with a Single Public Link
Notebooks can be shared publicly with a link, which is handy for publishing a research summary, a class study guide or a team briefing without handing out edit access to your source material.
Run Parallel Notebooks, Not One Mega-Notebook
Advanced users keep three to five active notebooks per project area instead of one for everything. A “competitor tracking” notebook, a “personal learning” notebook and a “current project” notebook, each tightly curated, will consistently outperform one sprawling notebook. It comes straight back to the curation principle in Module 3.
Ground Your Writing in a Notebook
In Google Docs, you can use a Gemini Notebook as a context source, so AI drafting is grounded in your curated sources without you switching tabs or copy-pasting. It’s one of the cleaner ways to go from research to a first draft.
Module 8: The Cloud Computer and the Gemini App (New in 2026)
Two changes since the rename deserve their own lesson, because they alter what a notebook can do.
A Secure Cloud Computer in Every Notebook
When Google renamed the tool, it also announced a secure cloud computer inside each notebook. It lets Gemini Notebook write and run code over your uploaded sources, for data analysis and for outputs that need real calculation instead of a text summary. It reached Ultra and eligible Workspace accounts first, and Google said it would expand to Pro users on the web, so check whether your account has it before you plan a workflow around it.
Good first uses: upload a spreadsheet or CSV and ask for trends, outliers and a short written readout; or give it several reports and ask it to compute and compare the numbers they cite. Verify the figures against the source, as you would with any analysis.
Notebooks Beyond the Standalone App
Notebooks now live in more than one place. You can create and use them in the Gemini app, with syncing between the app and the standalone Gemini Notebook, and Google says notebooks are coming to AI Mode in Search. Gemini Notebook remains its own product focused on research. The Gemini-app version is closer to a personal knowledge base for your files and chats.
What Google Added on September 15
Google’s September update focused on study and capture: real-time conversations with your notebook, audio recording from the mobile app, and interactive overviews that include quizzes, flashcards and video summaries. Features roll out in phases, beginning on the web, so don’t assume every one is on your phone yet.
Module 9: Four Complete Projects, Start to Finish
These are full walkthroughs, not just ideas. Follow any of them end to end as practice.
Project 1: Turn Research Papers Into a Literature Review + Slide Deck
- Create a new notebook. Upload 10-15 PDFs of papers on your topic.
- Run Deep Research from the sources panel: “Identify the major schools of thought across these papers and where they disagree.”
- Add that report back in as a source.
- In chat: “Draft a literature review structure organized by theme, not by paper.”
- Generate a Slide Deck, revise weak slides with feedback, export as PPTX and finish the formatting in PowerPoint.
Project 2: Build a Personal Knowledge Base From a YouTube Channel
- New notebook. Add 15-20 video links from a channel you learn from regularly.
- Generate a Mind Map to see how the channel’s core topics connect.
- Generate an Audio Overview with a focus prompt: “Summarize the three most repeated pieces of advice across these videos.”
- Generate a study-guide Report to keep as a permanent reference.
Project 3: Competitive Analysis for a Business Decision
- New notebook. Upload any internal documents you have on your product and pricing.
- Run Deep Research: “Research how [Competitor A], [Competitor B] and [Competitor C] position and price this feature, and summarize public customer complaints about each.”
- Follow up in chat: “Which competitor’s approach has the most independent criticism, and what exactly is criticized?”
- Turn the findings into a Data Table, export it to Sheets, and generate a Slide Deck to present to your team.
Project 4: Exam Prep Pipeline for Students
- New notebook. Upload your course readings, lecture slides and any lecture recordings (as audio or video sources).
- Generate a study-guide Report for a structured overview.
- Generate Flashcards and Quizzes, using the “Got it / Missed it” sort to focus review time.
- Generate an Audio Overview to review passively on your commute.
- Before the exam, ask in chat: “What are the three concepts I’m most likely to get wrong, based on how often they’re emphasized versus how nuanced they are?”
Module 10: Limitations, Quotas & Privacy – What to Expect
No course is complete without the honest limitations section.
- It can only know what you give it. That’s the core strength and the core limitation. It won’t reason about current events or facts outside your sources unless you use Deep Research to pull in outside material.
- Source caps are real. Each source tops out at 500,000 words or 200 MB, and Free allows 50 sources per notebook. Very large documents may need splitting. Sharing a notebook doesn’t raise anyone else’s source limit.
- Usage is a budget now, not a count. Heavy outputs eat it faster than chat does, and it can run out before the week ends if you use the tool around the clock. Check Settings → Usage.
- It’s cloud-only. Gemini Notebook needs an internet connection and processes your files on Google’s servers. If strict data locality is a requirement, that matters. Data-handling terms differ between personal and Workspace accounts, so read the privacy page for yours.
- Admins can turn it off. In a managed Workspace account, most of these features have no per-user toggle. If it’s missing at work, that’s likely an admin-level setting, not something you can switch on yourself.
- Limits and plan rules change. Google says access and limits are subject to change, which is why this page carries a “last updated” date.
Module 11: Quick-Reference Cheat Sheet
| Task | Where to Go |
|---|---|
| Start a new project | New Notebook → add 5-15 tightly relevant sources |
| Find more sources | Discover (up to about ten suggested web sources) in the sources panel |
| Multi-step investigation | Sources panel → Web → Deep Research (scope it as a question, not a fact lookup) |
| Passive review / commute learning | Audio Overview with a focus prompt |
| Present findings | Slide Deck → export as PPTX or PDF |
| Exam/test prep | Reports (Study Guide) + Flashcards + Quizzes |
| Analyze numbers in your sources | The notebook’s cloud computer, where your plan has it |
| Reuse a notebook structure | Copy Notebook |
| Check what you have left | Settings → Usage |
| Share results externally | Public share link |
FAQ
Does Google use my uploads to train its AI?
Not on personal accounts, with one exception. Google says your data isn’t used to train Gemini Notebook unless you submit feedback, in which case it may review the full context of that interaction, including your queries, uploads and the model’s replies. On Google Workspace and Workspace for Education accounts, nothing is reviewed by human reviewers or used to train models, even when you give a thumbs up or down.
Do Gemini Notebook videos and images carry a watermark, and can I remove it?
Everything made with Veo, Omni or Nano Banana carries an invisible SynthID watermark that you can’t switch off. A visible watermark is added too, and on Google AI Pro and Ultra you can turn it off with the Visible watermarking toggle under your profile picture. Users in India, South Korea and Vietnam always get the visible one.
Can students get Pro features for free?
Yes, through a relaunched offer rather than a discount. Since August 19, 2026, eligible US college students get 12 months of Google AI Pro free, which carries the Pro tier, and students in more than 140 other markets get 12 months of Google AI Plus. Redeem by December 31, 2026. Fello AI reports that a payment method is required and the plan auto-charges $19.99 a month afterwards, so set a reminder.
Can I use books I own as sources?
Some of them. Under Google’s Expert Intelligence initiative, you can add select ebooks you’ve bought on Google Play Books, from 100,000-plus titles by publishers such as Penguin Random House, Macmillan and O’Reilly. You must own the book, and collaborators on a shared notebook have to buy their own copy before they can query it. Look for Gemini Notebook under the Tools badge on a book’s Play Books page.
What happens if I run out of usage and want a big output anyway?
On the web, choose Generate later in the Studio panel and Gemini Notebook queues heavy jobs, such as Video Overviews or Slide Decks, and runs them automatically once your allowance resets, notifying you if notifications are on. It may take a couple of hours. Gemini Notebook may also suggest a lighter output when your first choice would exceed your limit. The option isn’t on mobile yet.
Can other people copy my notebook, and how do I stop them?
If you share a notebook, others can make a copy unless you switch it off. Google’s setting is called Allow copies in the notebook’s sharing options, and it’s on by default. A copy carries the sources and Studio items, but Drive-based sources come across only if the person copying has permission to access and copy the original Drive file. Turn the setting off before sharing sensitive material.
Related Reading on AI Tech Safar
- What Is Generative AI, Really? A Plain-English Guide (2026)
- Agentic AI Explained: What It Is, How It Works, and Why 2026 Is the Tipping Point
- How to Use ChatGPT Agent Mode (Now Called Work): The Complete Step-by-Step Guide (2026)
Useful Sources
- Google Blog – NotebookLM is now Gemini Notebook
- Google Help – Learn about Gemini Notebook’s plans
- Google Help – Manage your Gemini Notebook usage limits
- Google Workspace Updates – New ways to customize and interact with content (March 2026)
- Google Workspace Updates – Make a copy of a notebook in Gemini Notebook
- Google Workspace Updates – Automatically add sources in Workspace Studio
- Google Workspace Updates – Ground AI prompts in Google Docs with Gemini Notebook
- Google Blog – Expert Intelligence in Gemini Notebook
- 9to5Google – Gemini Notebook switching to compute-based usage limits
- Fello AI – NotebookLM Pricing 2026: Free vs Plus vs Pro vs Ultra
- Fello AI – NotebookLM Update 2026: Every New Feature Explained
- DEV Community – Google Expands Gemini Notebook With Syncing, Interactive Tools and a New Name
- Elephas – NotebookLM Daily Limits 2026: New 5-Hour Usage Limits Explained
For the broader context and related reading, explore our AI Tools 2026: Free Options, Categories and Companies.


