NotebookLM: Build Your Own Knowledge Base
Google's free tool that turns your documents into something you can question, study, listen to, and learn from.
If I could only recommend one AI tool to someone who thinks AI is not for them, it would be NotebookLM. It does something no other tool does as well: you give it your own documents, and it becomes an expert on that specific material. Not the whole internet. Just your stuff.
Upload a contract, a research paper, a set of meeting notes, a textbook chapter, or even a YouTube video. Then ask questions. NotebookLM answers using only what you gave it, and it cites the exact sources. No hallucinations from training data. No wandering off-topic. Just grounded answers from your material.
That alone would be useful. But NotebookLM goes much further.
| NotebookLM | |
|---|---|
| What it is | AI-powered research and study tool that analyses your uploaded sources |
| Best at | Deep analysis of your own documents, generating podcasts and study materials, building a personal knowledge base |
| Free tier | Free with a Google account: 100 notebooks, 50 sources per notebook, 3 audio overviews per day, 50 chat queries per day |
| First paid tier | NotebookLM Plus (bundled with Google AI Pro): US$20/month (~A$32/month). 500 notebooks, 300 sources per notebook, 20 audio overviews per day, 500 queries per day |
| Ken's take | The free tier is generous enough for most people. The audio overview feature alone makes it worth trying. I use it to build personal knowledge bases on topics I keep returning to, and as a prompt library I can search with natural language. It is the single most underrated AI tool available. |
| Sign up | notebooklm.google.com (Google account required) |
How It Works
The concept is simple. You create a notebook. You add sources to it. Then you use the chat and the Studio panel to do things with those sources.
A notebook is just a container for a topic. You might have one notebook called "House Renovation", another called "Work Project Q2", another called "Family History". Each notebook is separate. The AI in one notebook has no knowledge of what is in another.
Sources are the material you feed it. NotebookLM accepts Google Docs, Google Slides, PDFs, text files, Markdown files, web URLs, YouTube video URLs, audio files, images (with text recognition), CSV files, and EPUB books. You can paste text directly too. Each source can be up to 500,000 words or 200MB for uploaded files.
Once your sources are loaded, you can chat with them. Ask questions. Request summaries. Ask it to compare two sources. Ask it to find contradictions. Every answer comes with inline citations that link back to the specific passage in your source material, so you can verify anything it says.
The Studio Panel: Where It Gets Interesting
The chat is useful, but the Studio panel is where NotebookLM really pulls ahead. With one or two clicks, you can generate any of the following from your sources:
Audio and Video
Audio Overviews. This is the feature that made NotebookLM famous. Click a button and it generates a podcast-style conversation between two AI hosts who discuss your material. They summarise key points, make connections, explain difficult concepts in plain language, and even banter. The result is usually 6 to 15 minutes long, and it is genuinely engaging. You can download it and listen on the go. I have used this to absorb dense policy documents while walking the dog. There are multiple formats: Deep Dive (thorough exploration), Brief (quick summary), Critique (critical analysis), and Debate (two perspectives).
Interactive Mode. While listening to an audio overview, you can "raise your hand" and interrupt the hosts. Ask a follow-up question and the AI hosts answer using your sources, then pick up where they left off. This sounds gimmicky. It is not. Being able to stop a podcast and say "wait, explain that part about the timeline" is remarkably useful.
Video Overviews. New in early 2026, NotebookLM can generate animated video explainers from your sources. The basic version creates narrated slides. The Cinematic Video Overviews (available on the Ultra plan) use more advanced AI models to produce fluid animations with rich visuals. Useful for turning dense material into something you can share with people who will not read a 40-page document.
Visual and Structural Outputs
Mind Maps. Generates a visual branching diagram of the main topics and connections in your sources. Good for getting a quick structural overview of complex material.
Slide Decks. Creates a presentation from your sources that you can refine and edit. Not as polished as a hand-built deck, but a solid starting point.
Infographics. Generates visual summaries in a range of styles: Professional, Scientific, Sketch Note, Editorial, Bento Grid, and several others. You can choose landscape, portrait, or square format and adjust the level of detail.
Study and Reference Tools
Study Guides, Flashcards, and Quizzes. Upload your study material and NotebookLM can generate structured study guides, flashcard decks (with spaced repetition tracking), and multiple-choice quizzes. Progress saves across sessions. You can mark flashcards as "Got it" or "Missed it" and rerun the ones you struggled with. Useful for anyone studying, or for testing your understanding of professional material before a meeting or presentation.
Briefing Documents. Generates a structured summary of your sources in briefing format. Handy for preparing for meetings where you need to quickly absorb background material.
Data Tables. Pulls scattered information from your sources into structured tables you can sort and filter. Good for comparing information that is spread across multiple documents.
Timelines. Extracts events and dates from your sources and arranges them chronologically. Works well for historical material, project planning documents, and legal case files.
FAQs. Generates a set of questions and answers from your sources. Useful for creating quick reference guides or testing whether a document covers the questions your audience is likely to ask.
What to Actually Use It For
Here are the use cases I keep coming back to.
A prompt library you can search. Create a notebook called "My AI Prompts". Add your best prompts as text sources, grouped however you like. Then when you need a prompt, ask NotebookLM in natural language: "Find my prompts about writing cover letters" or "Which prompt did I use for summarising meeting notes?" This is more useful than a spreadsheet because you can search by meaning, not just keywords.
Studying complex documents. Upload a contract, a government policy paper, an insurance policy, or a body corporate report. Ask NotebookLM to summarise the key obligations, flag anything unusual, or explain specific clauses in plain language. Generate an audio overview and listen to it while you do something else. Come back and ask follow-up questions.
Meeting preparation. Dump all the relevant documents into a notebook: the agenda, last quarter's report, the project plan, the email thread that started the whole discussion. Ask NotebookLM to brief you. Generate a FAQ of likely questions. Walk into the meeting prepared.
Learning a new subject. Add a textbook chapter, a few key articles, and a relevant YouTube video to a notebook. Generate an audio overview to get oriented. Then use the chat to ask questions as you work through the material. Generate flashcards to test yourself later.
Family history and genealogy. Upload interview transcripts, scanned documents, family letters, and historical records. Ask NotebookLM to find connections, build timelines, and identify gaps in your research. Generate an audio overview to share with family members who would rather listen than read.
Travel planning. Add guidebook pages, hotel booking confirmations, flight itineraries, and blog posts about your destination. Ask NotebookLM to build a day-by-day plan, flag logistics issues, or suggest restaurants near your hotel based on the sources you gave it.
Getting Started in Five Minutes
Go to notebooklm.google.com and sign in with your Google account. Click New Notebook. Give it a name. Then add sources using the upload button: drag in a PDF, paste a URL, or connect a Google Doc.
Once your sources are loaded, try asking a question in the chat. Then open the Studio panel (on the right side of the screen) and click Audio Overview. Wait a couple of minutes. Press play. That is usually the moment people understand why this tool is worth their time.
Try this right now (free)
Go to NotebookLM, create a new notebook, and paste in this URL as a source: pick any long Wikipedia article on a topic you are curious about, or a YouTube video URL of a talk you have been meaning to watch. Then generate an Audio Overview. Listen to the first few minutes. You will immediately see why people are excited about this tool.
Pricing in Detail
The free tier is genuinely useful. For most people experimenting with NotebookLM, it is enough. Here is how the tiers break down:
| Feature | Free | Plus (US$20/month, ~A$32) |
|---|---|---|
| Notebooks | 100 | 500 |
| Sources per notebook | 50 | 300 |
| Chat queries per day | 50 | 500 |
| Audio overviews per day | 3 | 20 |
| Early access to new features | No | Yes |
NotebookLM Plus is bundled with the Google AI Pro subscription (US$20/month, ~A$32/month), which also gets you Gemini embedded in Gmail, Docs, Sheets, and Slides, plus access to Gemini 3 and Deep Research. If you already pay for Google AI Pro, you already have NotebookLM Plus. There is also an Ultra tier at US$250/month (~A$400) aimed at power users and businesses, which adds features like Cinematic Video Overviews and 600 sources per notebook. Most people do not need Ultra.
Students in the US can get Google AI Pro at a 50% discount (US$10/month).
What You Can Upload
NotebookLM accepts the following source types:
Documents: Google Docs, PDFs, text files (.txt), Markdown files (.md), EPUB books.
Presentations: Google Slides.
Data: CSV files.
Media: Audio files, images (with optical character recognition for text in photos).
Web content: Any public URL, YouTube video URLs (it extracts and analyses the transcript).
Direct input: Copy and paste text directly as a source.
Each source can be up to 500,000 words or 200MB. On the free plan, you can have up to 50 sources per notebook. That is a lot of material.
Tips From Experience
Be specific with your notebook topics. A notebook called "Everything About Work" will be less useful than three notebooks called "Q2 Product Launch", "HR Policy Questions", and "Client Meeting Notes". Narrower topics produce better answers.
Name your sources clearly. When you have 20 sources in a notebook, you want to know which is which at a glance. "Board Minutes - March 2026" is better than "document-final-v3.pdf".
Use the citation links. When NotebookLM gives you an answer, it includes numbered citations. Click them. They take you to the exact passage in your source material. This is how you verify that the AI is not extrapolating beyond what the sources actually say.
Combine source types. The best notebooks mix source types. A YouTube video of a conference talk, the speaker's published paper, and a blog post analysing the same topic. NotebookLM will synthesise across all of them.
Revisit and add sources over time. A notebook is not a one-shot thing. Start with a few sources. Use it. Find new material. Add it. Your notebook gets smarter as your collection grows.
Download your audio overviews. The generated podcasts are surprisingly good for sharing with colleagues, family members, or anyone who prefers listening to reading. Save them and share them.