NotebookLM just got a massive upgrade inside Gemini

Yesterday I stumbled on a walkthrough that made me rethink how I’ve been using NotebookLM. Turns out I was barely scratching the surface. The creator behind this breakdown, Learnit Training, has been using NotebookLM for nearly 3 years and just dropped 10 tips that honestly made me feel like a beginner again.

Here’s the twist: the biggest update isn’t even inside NotebookLM itself. Google quietly rolled notebooks directly into Gemini’s interface, which changes the whole workflow.

🔄 What’s New

NotebookLM now lives inside Google Gemini. If you go to gemini.google.com, there’s a new “Notebooks” section on the left sidebar. You can create notebooks, add sources, and chat with them without ever leaving Gemini. Previously you had to bounce between NotebookLM’s own site and Gemini, attaching notebooks manually. That friction is gone.

But that’s just the starting point. The expert also walks through a research-first approach that flips how most people use the tool. Instead of uploading your own files right away, you let NotebookLM find sources for you using two modes: fast research (grabs about 10 web sources quickly) and deep research (an in-depth report that takes up to 10 minutes but pulls way more material). The smart move is to run both, cherry-pick the best sources, then layer in your own files last. With a paid plan, you can stack up to 300 sources per notebook. Free users get 50.

🧩 The Twist: Notes Become Sources

This is the part that caught me off guard. When you’re chatting with your notebook and get a really strong answer, you can save it as a note. Nothing new there. But then you can convert that note into a source. That means your best AI-generated insights feed back into the notebook as reference material, making future answers sharper and faster.

The workflow looks like this:

1. Chat with your sources and get a solid answer
2. Save that answer as a note
3. Convert the note to a source
4. Now the notebook references that refined insight alongside your original material

You can also create blank notes, paste in your own thoughts or snippets from the chat, format them however you want, and eventually promote those to sources too. It’s like building a knowledge base inside the notebook as you go.

📋 Step-by-Step Mini-Workflow

Here’s the approach the creator recommends for getting the most out of NotebookLM on any project:

1. Start with fast research on your topic using the web search option. Review each source individually before importing. Don’t just dump everything in.
2. Run deep research on the same topic. This takes longer but surfaces sources you’d never find manually. Add the best ones.
3. Layer in your own files as the final step: PDFs, Google Drive docs, website URLs, or pasted text.
4. Talk to individual sources by deselecting everything and clicking just one. This gives faster, more specific answers when you have dozens of sources loaded.
5. Use mind maps first when a topic feels messy. Instead of jumping straight into chat, generate a mind map to see how your sources connect visually. It’s downloadable as a PNG.
6. Generate infographics from individual sources or your entire notebook using NotebookLM Studio. Multiple styles are available (professional, editorial, instructional) and you control the detail level.
7. Create multiple audio overviews for different audiences. You can generate a technical deep dive, a beginner-friendly brief version, and even versions in different languages. They all stack up next to each other.

🎬 Cinematic Video Overview

This is the newest feature the creator highlights, and it’s only available on paid plans right now. Video overview goes beyond the familiar two-host podcast format. The cinematic mode produces something that actually looks and feels unique each time. You can customize it with prompts, tailor it for specific audiences, or let it run on defaults. The explainer and brief video modes are available on free plans.

💡 Pro Tips

• Interactive mode on audio overviews is wildly underused. After generating a podcast, you can actually interrupt the AI hosts mid-conversation, ask a question, get an answer, and then they continue where they left off. It turns a passive listen into an active tutoring session.
• Custom notebook settings let you upload a cover image and override the auto-generated summary with your own. This matters when you’re sharing notebooks with teammates or clients.
• Chat customization (paid plan) lets you set a system prompt that changes how Gemini responds inside that specific notebook. Want shorter answers? More technical depth? You define the conversational style once and it sticks.
• Sharing controls give you options for link access, public visibility, or restricted invite-only access per notebook.

The real value here isn’t any single feature. It’s how they chain together. Research finds your sources, notes capture your best insights, those notes become sources, and then Studio turns everything into visuals, audio, or video you can actually share. The whole thing compounds.

Check out the full video for detailed walkthroughs of each feature, especially the interactive mode demo and cinematic video output. Worth watching if you use NotebookLM for any kind of research or content work.