You've got a pile of PDFs to read, YouTube videos to watch, articles to digest — and not enough hours in the day. That's exactly why I started testing NotebookLM, now called Gemini Notebook since July 2026. This free AI tool from Google takes your documents, analyzes them, and spits out summaries, flashcards, quizzes, and even a two-voice podcast. I'll walk you through how it works, what it's actually worth, and how to use it without writing a single line of code.
NotebookLM is a name you may have heard floating around in 2024 or 2025. Since July 2026, Google rebranded it as Gemini Notebook — a name change that reflects its deeper integration into the Gemini ecosystem. [F1]
The concept is easy to grasp. You load documents into a "notebook." The tool reads them, indexes them, and becomes your personal assistant for that specific set of content. You can ask it questions, request a summary, have it generate study flashcards, a quiz — or even a two-voice podcast that discusses what's in your sources.
What sets it apart from using ChatGPT or Claude on their own is that Gemini Notebook only answers based on what you've given it. It doesn't go browsing the internet at random. It stays inside your notebook. Technically, this is called RAG — Retrieval Augmented Generation — a method that pulls specific passages from your sources before forming a response. [F5] In practice, that means fewer hallucinations, fewer made-up facts, and answers you can verify in one click since every claim is linked to a passage in your documents.
It works with any personal Google account. Free by default, with advanced features available on a Pro plan. [F2] For an entrepreneur who wants to try it without spending a dime, it's a solid starting point.
If you want to dig deeper into the tool's definition and history, I have a dedicated entry in the NotebookLM glossary page.
The output formats available in Gemini Notebook are more varied than you'd expect at first glance. [F3]
Summaries and answers. That's the foundation. You load an 80-page PDF, ask for a 10-point summary, and you've got it in seconds. You can also ask targeted questions: "What are the recommendations in chapter 4?" or "Is there any mention of delivery timelines?" The tool answers and cites the exact passage.
Study guides, flashcards, and quizzes. Useful when you're learning something new — a regulation, an industry, a tool. Load your sources, ask for a set of flashcards or a multiple-choice quiz. This has been available in English since the tool launched, and French support arrived in April 2025. [F3]
Audio Overviews — the famous podcast feature. This is what made the tool go viral. Gemini Notebook generates a conversation between two synthetic voices that discuss your documents like a real podcast. The result is surprisingly smooth. You can listen in the car, while working out, while cooking. Since a March 2026 update, there's also a "Conference Mode" and the ability to clone a voice for these podcasts. [F10]
Animated educational videos. More recent: Gemini Notebook can now generate animated videos (called Video Overviews, powered by Google's VEO technology) from your sources. [F10] It's still an emerging feature, but the idea of turning a report into a short explainer video is genuinely interesting.
Export and web search. Since the "Agentic Upgrade" update in June 2026, the tool integrates Gemini 3.5, a "from scratch" web search (to build a notebook from an online search rather than existing documents), and export via the Studio Panel. [F4]
For a deeper dive into summarizing PDFs and videos with AI, I wrote a dedicated article: how to summarize a PDF or video with AI.
The two-voice podcast is genuinely impressive the first time you hear it. But what's won me over day-to-day is the simplicity of sourced answers. You ask a question about a 40-page contract, the tool answers and shows you the exact paragraph. No more spending 20 minutes hunting for one sentence.
48% of French people used generative AI in 2026, up from 20% in 2023 — a doubling in three years. [S1] And 47% of French SMEs had launched at least one AI project by March 2026, up 18 points in twelve months. [S3] These numbers aren't here to impress — they show that AI is no longer a topic reserved for tech teams. It's becoming part of everyday practice for entrepreneurs, including those who don't write code.
Gemini Notebook fits exactly into that logic. No API to configure, no tools to connect, no code to write. You open a browser, create a notebook, load your files, and get to work.
The use cases I see most often:
The tool doesn't replace careful reading when that's truly necessary. But it cuts down the time you spend hunting for information in a document you already know 60% of.
If you want to get better at phrasing your requests to the tool, that's a prompting question. I have a practical guide on how to write a good prompt when you're not a developer.
The tool has evolved a lot over the past few months. Here are the points I think matter most if you're starting now.
The name change. NotebookLM became Gemini Notebook in July 2026. [F1] If you search for the tool on Google, both names take you to the same place. But going forward, "Gemini Notebook" is the official name.
Gemini 3.5 integration. The "Agentic Upgrade" update in June 2026 brought a more powerful engine, the ability to run code in a sandboxed environment, and better result export. [F4] In practice, answers are more precise and complex documents are handled more reliably.
Full language support. Audio Overviews in French have been available since April 2025. [F3] That was a major barrier to wider adoption — it's gone now.
The underlying technology. Gemini Notebook runs on Google's Gemini models. If you want to understand what powers the tool, I have a glossary entry on Gemini and its models.
Don't load dozens of documents without any organization. Create one notebook per project or topic. A "competitor watch" notebook, a "client X file" notebook, a "training Y" notebook. The more focused your notebook, the more relevant the answers.
Here's how I'd approach it if I were starting today.
Step 1 — Open a notebook. Go to notebooklm.google.com (the domain name hadn't changed at the time of writing). Sign in with your Google account. Click "New notebook."
Step 2 — Load your sources. The tool accepts PDFs, articles via URL, YouTube videos, and audio files. [F3] Start with one or two documents on a topic you know well — that way you can verify the quality of the answers.
Step 3 — Ask simple questions. "Summarize this document in 5 points." "What are the main recommendations?" "Are there any key figures?" Watch how the tool cites its sources. That traceability mechanism is what lets you trust (or question) each answer.
Step 4 — Generate an Audio Overview. Click the dedicated button. Let the tool generate the podcast. Listen to the first two minutes. You'll immediately understand why it's useful for long documents.
Step 5 — Create flashcards or a quiz. Ask it to generate flashcards on the key points. Test yourself. It's a quick way to check what actually stuck.
For more on automating your tasks without code, the article on automating your tasks with AI without writing code is a good follow-up.
Yes, basic access is free with any personal Google account, including Google Workspace accounts. [F2] Advanced features and higher usage limits are available on a paid Pro plan. For moderate daily use, the free version is more than enough to get started.
The tool accepts PDFs, articles via URL, YouTube videos, and audio files. [F3] So you can load an annual report, a blog post, a recorded conference, or a meeting recording. The range of supported formats is a real advantage.
The main difference is how they work. Gemini Notebook only answers based on the sources you provide, thanks to the RAG mechanism. [F5] ChatGPT and Claude, used without a search tool enabled, can answer from their general training — which is useful for open-ended questions, but less reliable for analyzing a specific document. For document research on a defined set of content, Gemini Notebook is more traceable. For open-ended conversation or content generation, ChatGPT or Claude can be complementary. If you want to compare models on other dimensions, I have an article on free vs. paid AI: how to choose.
Yes — all main features, including Audio Overviews (podcasts), are fully available in English. French support arrived in April 2025. [F3]
The free version has limits on the number of sources per notebook and the volume of requests. [F2] These limits aren't published as fixed numbers and may change over time. If you're working on intensive projects with lots of documents, it may be worth looking at the Pro plan. Either way, organizing your notebooks by topic will help you stay within reasonable volumes.
Every answer from Gemini Notebook is linked to a specific passage in your sources. [F5] You can click the reference to check the original excerpt. That's the best habit to build: don't just read the answer — verify the cited source. The tool reduces the risk of hallucinations compared to a general-purpose model, but it doesn't eliminate them entirely. Your judgment still matters.
Gemini Notebook is one of the first AI tools I recommend to anyone who wants to save time reading and synthesizing documents, without having to learn anything technical. It's free, it works in English, and its cited-source mechanism makes it more reliable than most alternatives for this specific use case.
It's not a magic tool. It doesn't replace your thinking, it doesn't read between the lines, and it won't understand your business context unless you give it the right documents. But for turning a pile of PDFs into usable information in a few minutes, I haven't found anything better so far.
I do all of this for myself first — and if it helps you too, all the better.
I test AI for real and share what actually works, no jargon, no hype. If this article was useful, the easiest way to stay in the loop is my Friday newsletter. And if you have a question or a doubt: reply to me, I read everything.