NotebookLM: How to Use Google's AI for Studying

Section 01What NotebookLM is and why it's now Gemini Notebook

NotebookLM is a notebook where you drop your materials, and Google's AI reads them and helps you work with them. A regular chatbot like Gemini or ChatGPT answers from its general knowledge and sometimes confidently invents things. NotebookLM works differently: it answers from the files you gave it and puts a small number next to each point. Click the number and you land on the exact spot in the lecture or textbook the answer came from.

For studying, that's the whole point. When you're prepping for an exam, you need answers from your textbook and your professor's lectures. A summary from the internet only muddies things. If the answer isn't in your materials, NotebookLM tells you so. It won't fill in the gaps.

The service launched in 2023 as an experimental Google Labs project. It took off in fall 2024, when Google added audio overviews: two AI hosts discuss your material like a podcast. Video overviews, mind maps, flashcards, quizzes, infographics, and slide decks came later.

What happened in summer 2026

On July 16, 2026, Google announced that NotebookLM is now called Gemini Notebook. It's a rename, and nothing was shut down: notebooks, files, and shared links work as before. Since late July the service has a second address, notebook.google, and the old notebooklm.google redirects there. The mobile app got a new name and icon too; just update it.

Along with the new name, notebooks started syncing with the Gemini app, and Google began adding an environment inside notebooks where the AI can run calculations on your spreadsheets by itself. Ultra subscribers got it first, then Pro. For studying it's a nice extra for now.

People still search for "NotebookLM," so I use the old name in this article. If you see "Gemini Notebook" in the interface, that's the same thing.

Section 02What NotebookLM can do

The NotebookLM screen has three parts. On the left are your sources, meaning your materials. In the middle is the chat. On the right is the Studio: buttons that turn your sources into ready-made study materials.

What you can upload

According to Google's help pages, you can add to a notebook:

• Files: PDF, Word and other Microsoft Office files, ePub, CSV, images, MP3 and WAV audio.
• Google Docs, Sheets, and Slides straight from Drive. Slide decks are capped at 100 slides.
• Website links. An article, a textbook page, documentation.
• YouTube videos. Only public videos with speech and captions, even auto-generated ones. A video uploaded less than three days ago may not go through.
• Pasted text. Paste notes from your notes app, and they become a source.

One source can be up to 500,000 words or 200 MB. That's roughly several thick textbooks in one file; in practice almost nobody hits that limit.

If you don't have much material, there's a source search button: you type a topic, and NotebookLM finds articles on the web or documents in your Google Drive and offers to add them to the notebook. Handy when you start a paper from scratch. Still, check what it finds with your own eyes, like any list of search results.

Chat with your materials

In the chat you ask in plain words: "explain chapter two in simple terms," "what's the difference between these two theories," "which dates do I need to remember." The answer comes with numbered citations. You can uncheck some sources on the left, and the AI will answer only from the selected ones. Useful when the notebook holds a whole semester but your question is about one lecture.

A good answer saves to a note with one button. You can later turn a note into a new source and ask questions about it.

Studio: what you get in one click

• Audio overview. Two voices discuss your material like podcast hosts. You can pick a format: deep dive, brief, critique, or a debate between two points of view. Great for commutes and walks.
• Video overview. A video with slides and a voiceover based on your sources. Some styles may not be available in every language yet.
• Mind map. A diagram of the topic with branches. Click a branch, and an explanation appears in the chat.
• Reports. A study guide with key terms and questions, a short summary, a timeline, an FAQ, a briefing doc.
• Flashcards. A question on the front, the answer on the back. The classic for drilling terms and dates.
• Quiz. Multiple-choice questions with an explanation of why the right answer is right.
• Infographics and slides. A cheat-sheet image and a deck based on your material. Useful for a presentation.

Every Studio button has a pencil icon for customization. There you can say what to focus on, the difficulty level, how many cards, and which language. More on that in the exam section.

Section 03NotebookLM free plan and pricing

NotebookLM is free. For studying, the free version is enough for almost everyone: a semester notebook with lectures, a textbook, and a study guide fits with room to spare. A paid plan raises the limits, and it's one subscription for all of Google's AI services.

The notebook and source counts were checked against the official Gemini Notebook plans page on November 5, 2026. Prices are US prices; on the Google AI plans page they may differ in your country, and tax can be added. Besides notebooks, a subscription includes Gemini, Drive storage, and other Google services.

How usage is counted now

Before September 2, 2026, it was simple: on the free plan you got 50 chat questions a day and 3 audio overviews a day. Now Google counts total compute. A heavy request across a hundred sources costs more, a short question about one lecture costs less, and a video overview costs more than flashcards. Your quota refreshes every 5 hours until you hit the weekly cap. Paid plans multiply that quota.

What it feels like in practice: if you ask questions on one topic and generate a couple of overviews in an evening, you'll rarely hit the limit. If you run video overviews for the whole semester back to back, the quota runs out faster, and you'll wait a couple of hours.

One more thing: when you first upload sources, NotebookLM sometimes builds a report, flashcards, or an audio overview on its own. According to Google's help page, those automatic materials don't count toward your quota.

Student offers

Google periodically runs student promotions on its AI plans, sometimes with Plus or Pro free for a period. Terms and eligible countries change, so check the Google AI plans page for current student offers.

Section 04Where NotebookLM works: countries, age, and account requirements

According to Google's help page, Gemini Notebook is available in the regions where the Gemini app is available. That list covers the US, Canada, the UK, the EU, and many other countries. If you're in the US, you can open it and start right away.

What you need to get in:

• A Google account. A regular Gmail account is enough for the free plan. No payment card is needed to start.
• Age requirement. Personal accounts must meet Google's minimum age for Gemini in your country. If your age isn't verified, the service may ask you to confirm it.
• School or work accounts. With a Google Workspace account (a university or company address), the administrator decides whether NotebookLM is turned on. If it won't open, ask your IT team or switch to a personal account.
• Keep your originals. Don't make NotebookLM the only place your study materials live. Keep the source files on your computer or Drive, so you lose nothing if access changes.
• If it isn't available where you are, or you'd rather stay in one tool, you can make the same summary, flashcards, and quiz in Claude. Section 10 shows how: I ran the prompt myself and included the full answer.

Section 05Your first NotebookLM notebook in 10 minutes

Grab material you actually need to learn: lectures for one course, a textbook chapter, an exam study guide. You'll figure things out faster on something real than on a demo.

That's it. From now on you have a personal tutor for your lectures that never gets tired of answering the same thing and always shows where it's written.

Section 06Study prompts: what to type in the NotebookLM chat

You can write in plain words, but a few phrasings work noticeably better than others. Here are the ones worth saving. Copy them and replace what's in square brackets.

The most useful one here is "test yourself." When the AI asks and you answer in your own words, the material sticks far better than rereading. In this mode NotebookLM acts like a patient tutor: if you get it wrong, it shows you the quote where the right answer is.

How to set up the chat for yourself

Chat settings include a conversation goal and response length. You can pick a preset, like a learning guide, or write your own instruction. Here's mine for exam prep:

The instruction applies to the whole notebook, so you don't have to type it every time.

Section 07Audio overview: how to turn a lecture into a podcast

The audio overview is the feature people fell in love with NotebookLM for. The AI takes your sources and builds a conversation between two hosts: one explains, the other asks follow-ups, reacts, and gives examples. It's easier to listen to than a straight reading, and more of it stays in your head.

How to make one

1. Check the sources you want the overview to cover. One or two lectures work best: the narrower the topic, the deeper the conversation.
2. In the Studio, click the pencil next to "Audio overview."
3. Pick a format. Deep dive for a new topic. Brief to refresh your memory. Critique and debate when you need to see a theory's strengths and weaknesses.
4. Choose the language and length, if offered.
5. In the instructions field, say what to focus on.
6. Click "Generate" and wait a few minutes. You can close the tab; the finished overview will appear in the Studio.

There's an interactive mode too: while listening, you can "join" the conversation and ask the hosts a question out loud. They answer from your sources and get back to the topic.

You can download the overview as a file and listen in the car, on the subway, on a run. An hour of commuting a day adds up to several topics a week that used to just slip away.

There's a downside: the hosts sometimes oversimplify and can get a subtle point slightly wrong. So use audio for the big picture and review. Check definitions and formulas in the chat text, where you get citations to the source.

Section 08Exam prep in NotebookLM in one evening

Here's the order that works when the exam is close. Every feature I covered above comes together here.

If the exam is oral, add one more step: ask for a 3-minute answer outline for each question and say it out loud. AI won't replace your memory, but it will help you gather and organize the material in one evening.

Four more ways to use it for learning

• A video course. Add links to the course's YouTube videos that have captions. Ask "in which video did they explain [topic]" instead of scrubbing through an hour-long recording.
• Thesis and term papers. Upload the articles and books from your reading list. Ask: "Which sources support the claim [your claim], and which argue against it? Give quotes." The citations help you find the page for a footnote fast. Write the paper itself yourself: schools require it, and that's how you'll be able to defend it.
• A foreign language. Upload a text or video in the language you're learning and ask for a list of unfamiliar words with translations and examples from the same text, then turn them into flashcards.
• Professional development. Put manuals and regulations into one notebook. Questions like "what changed in [document] compared with [previous version]" save hours.

If you record lectures on your phone, turn the recording into text first. I covered how to do it fast in the article on AI transcription. And how to make good notes with any AI is in a separate piece on AI for note-taking.

Section 09NotebookLM for work: if you sell services and courses

Students aren't the only ones who have to learn. If you sell courses, consulting, or services, you've probably built up a mountain of material: livestream recordings, lessons, notes, client conversations. NotebookLM helps you sort through it.

• Audit your own course. Upload lesson transcripts and ask: "What questions does a client still have after this course? Where do I repeat myself? Which steps are missing?" The cited answer shows which lesson has the problem.
• Collect frequent questions. Upload an export of questions from your support chat or comments. Ask it to group them by topic and count which ones repeat most. That gives you an FAQ section for your site and post topics.
• Make a lead magnet. Build a study guide or infographic from your own livestream recordings and polish it into a free resource. How to take that to a finished PDF is in my article on a lead magnet with AI in an hour.
• Research your niche. Use source search to collect articles on your niche and ask what people argue about and which questions remain open. Those are ready-made content topics.

One warning. Don't upload clients' personal data to a notebook: phone numbers, medical details, ID numbers. Google says it doesn't use your files for training unless you send feedback. But if you do send feedback, human reviewers may see the whole conversation, including uploaded files. Other people's data doesn't belong there.

Section 10Alternative if you need Claude

If NotebookLM isn't available to you, or you already work in Claude, Claude comes closest in convenience. It handles your files very well: you can upload PDFs and documents, and in Claude Projects you can collect a whole course's materials, much like a notebook. It won't make a podcast, but a summary, flashcards, a quiz, and a review of mistakes are easy. Go to claude.ai and sign in. Where to start is in Claude from scratch.

Here's a prompt that replaces the NotebookLM Studio. Upload a lecture to the chat or paste its text, then add:

I ran this prompt myself on a short excerpt from a study text about the sales funnel. The excerpt: "A sales funnel is a person's path from first hearing about you to buying. It usually has these stages: attention, interest, desire, action. This model is called AIDA. At each stage some people drop off, which is why the funnel is drawn narrowing. To see where people are lost, you measure conversion, the share of people who moved on to the next stage." Here's the full answer, unedited:

What's good here: every flashcard answer is backed by a quote, so you can see right away that the AI didn't make anything up. In the last line, Claude itself pointed out what isn't in the text. That's the same "sources only" behavior people value NotebookLM for. On a long lecture, ask for 20 to 30 flashcards and a 10-question quiz; the format stays the same.

Which AI to pick for study and work overall, I compared in detail in Claude vs ChatGPT. And if you need web search with sources, look at Perplexity.

Section 11If something goes wrong

Section 12The whole path in one minute