LaunchpadHQ

Gemini Notebook

by Google

Gemini Notebook was formerly NotebookLM

Google renamed NotebookLM to Gemini Notebook on July 16, 2026.

If you used it: Nothing to do. Old NotebookLM links redirect to the new address, and our getting-started guide still uses the NotebookLM name in places.

Source: blog.google · checked 2026-10-20

Getting-started guide

Getting Started with NotebookLM

Upload your own documents and NotebookLM answers questions using only those sources — with citations, and even a podcast-style audio summary.

Current prices: Free · See all plans

Last reviewed 2026-07-13 · for Gemini Notebook

How it works

You create a notebook, add your sources — files, links, notes — and then chat with them: ask for a summary, a comparison, an explanation, and it answers from those documents only, with citations you click to see the passage. Add another source and ask again; the answers change with what's in the notebook.

What it is

NotebookLM is Google's source-grounded research assistant. You upload your own material — PDFs, Google Docs, web pages, YouTube videos, notes — and it answers questions using only those sources, citing exactly where each answer came from. Because it's anchored to what you gave it, it rarely makes things up. It also turns your sources into study aids: an Audio Overview (a podcast-style conversation about your material), video overviews, mind maps, flashcards, and summaries. It's for students, researchers, and anyone making sense of a stack of documents.

Before you start

Free with a Google account, no time limit: up to 100 notebooks, 50 sources each, and a daily chat allowance that covers a real study session, plus a few Audio Overviews a day. Most people never need to pay.

Paid tiers (bundled into Google's AI plans, which start at a low monthly price) mainly raise those limits and add heavier features.

One structural thing to know: it's cloud-only (no offline mode), and every source caps at a large but real size limit — a single enormous file may not import.

Security & privacy

NotebookLM's defining trait is that it answers only from the sources you upload, with citations — so unlike an open chatbot, it rarely fabricates, and you can trace every answer back to a specific document. That grounding is the whole point.

Where your documents stand

  • Google states it does not train its models on the documents you upload, and your sources stay private unless you choose to share a notebook.
  • But your files are processed on Google's servers (it's cloud-only) — so for truly confidential or regulated material, the personal version isn't the right home; that's what the enterprise version exists for.
  • Each notebook is self-contained — it can't pull from your other notebooks or the open web, which is a privacy feature as much as a limitation.

It's grounded, not infallible: it can still misread a source or miss context. Click the citations to confirm, exactly as you would with any research tool.

What it's good (and not good) at

Best at

  • Answering questions from your own documents, with citations
  • Turning a pile of PDFs into a searchable, summarizable knowledge base
  • Making study aids — audio overviews, mind maps, flashcards — from your sources
  • Staying grounded, so it rarely invents facts

Okay at

  • Very large single files (each source has a size ceiling)
  • Working across many notebooks at once (each is siloed)
  • Confidential/regulated material (cloud-only; use the enterprise version)

Avoid

  • Uploading sensitive files that shouldn't leave your control
  • Treating the audio/summary as verified without checking citations
  • Expecting it to answer from outside your uploaded sources

Starter prompts to try

Make a notebook and add sources

Create a notebook and upload a few related documents — a PDF, a Google Doc, a web article on one topic.

Why this one: Everything grounds on your sources; this is the step that makes NotebookLM different from a chatbot.

Tweak: Start with material you already need to understand — lecture notes, a report, a contract.

Ask a grounded question

Summarize the main argument across these sources, and list the points where they disagree. Cite where each point comes from.

Why this one: Shows the core value — a cited synthesis of your material, not the open web.

Tweak: Follow up by asking it to explain any point "like I'm new to this."

Generate an Audio Overview

Create the podcast-style Audio Overview and listen to two AI hosts discuss your sources.

Why this one: It's the feature that made NotebookLM famous, and a genuinely different way to absorb dense material.

Tweak: Great for reviewing on a commute — but still check the citations for anything you'll rely on.

Common pitfalls

  • It only knows what you give it. That's the point — it won't answer from the open web, so add the right sources.
  • Cloud-only, files on Google's servers. Fine for most material; not for truly confidential or regulated files (use the enterprise version).
  • Grounded isn't infallible. It can still misread a source — click citations before relying on an answer.

Where to next

Want the full picture first? See the Gemini Notebook overview for pricing, the free tier and related tools.