How to Use NotebookLM: A Step-by-Step Guide

To use NotebookLM, sign in with a Google account, create a notebook, add sources like PDFs or Google Docs, then ask questions or generate an audio overview grounded only in those sources. The whole setup takes about five minutes.

NotebookLM is Google's research tool built on Gemini. It reads whatever documents you give it and answers questions using only that material, instead of guessing from general web knowledge. That makes it useful for studying a textbook, prepping for a meeting, or turning a stack of research into a podcast-style summary.

This guide walks through every step: creating your first notebook, adding sources from PDFs, Google Docs, YouTube, and the web, generating an audio overview, and setting share permissions. It also covers the source limits people hit without warning.

How Do You Sign Up for NotebookLM?

You sign up for NotebookLM with any personal Google account at notebooklm.google, no separate account or credit card needed.

Go to the site and click Sign In. If you already use Gmail, Google Docs, or Google Drive, the same login works. Work or school Google accounts sometimes need an administrator to turn NotebookLM on first, so if you hit a blocked screen there, ask your IT admin rather than troubleshooting on your own.

NotebookLM is free to start. Google also sells NotebookLM Business and higher usage limits through its Google AI Plus, Pro, and Ultra subscription plans, which raise the notebook count, source count, and daily audio generation caps beyond the free tier.

• Works with any personal Google account, no separate sign-up
• Free tier included automatically once signed in
• Work/school accounts may need admin approval first

How Do You Create Your First Notebook?

You create a notebook by clicking Create New on the NotebookLM home screen, which opens an empty workspace waiting for sources.

Every notebook starts blank on purpose. NotebookLM will not answer questions until you feed it something to read. Think of a notebook as one project folder: one for a class, one for a client, one for a book you're studying. Keeping topics in separate notebooks keeps its answers focused, because NotebookLM only searches the sources inside the notebook you're currently viewing.

Name the notebook something specific right away. NotebookLM auto-generates a title from your first source, but a generic name gets confusing fast once you have a dozen notebooks running.

• Click "Create New" to open a blank workspace
• One notebook per project or topic works best
• Rename it immediately instead of keeping the auto-generated title

How Do You Add Sources to NotebookLM?

You add sources by clicking Add Source inside a notebook, then uploading a file, pasting a link, or pasting text directly.

NotebookLM accepts PDFs, Google Docs, Microsoft Word files, plain text, Markdown, ePub, Google Slides, PowerPoint files, Google Sheets and CSV files, audio files, images, YouTube links, and general web URLs. For YouTube, the video must be public and have captions, since NotebookLM pulls the transcript rather than watching the video. For web pages, it only captures the text, not images or embedded widgets, and it cannot open pages behind a paywall or login.

You can add sources one at a time or in a batch. Google Drive sources stay linked and update automatically if the original file changes, though if you lose access to that Drive file, it still counts against your source limit even though NotebookLM can no longer read it.

• Documents: PDF, Google Docs, Word, TXT, Markdown, ePub
• Media: YouTube (public, with captions), audio files, images
• Web: direct URLs (text only) and pasted text
• Spreadsheets/slides: Google Sheets/CSV, Google Slides/PowerPoint

What Are NotebookLM Source Limits?

NotebookLM's free tier caps you at 50 sources per notebook, with each source holding up to 500,000 words or up to 200MB for an uploaded file.

That word cap covers most single documents, including entire textbooks, but a 50-source ceiling matters more than people expect. If you're aggregating a full research library, hitting 50 sources happens faster than it sounds, especially when each web page or PDF chapter counts as its own source.

The free plan also limits daily activity: roughly 50 chat questions and 3 audio overview generations per day, and up to 100 notebooks total. Paid Google AI plans raise these caps, but check the current numbers on the vendor's pricing page before budgeting around a specific figure, since Google has adjusted these limits before.

• Free tier: 50 sources per notebook, 500,000 words or 200MB per source
• Free tier: about 50 chat queries and 3 audio overviews per day
• Up to 100 notebooks per account on the free tier
• Paid Google AI plans raise these limits — verify current numbers on notebooklm.google/plans

How Do You Ask NotebookLM Questions About Your Sources?

You ask questions by typing into the chat box at the bottom of the notebook, and NotebookLM answers only from the sources you added.

Every answer comes with inline citations pointing back to the exact source and passage it used. Click a citation and NotebookLM jumps to that spot in the original document, which makes it easy to check the answer instead of taking it on faith. This citation habit is the main reason NotebookLM feels more trustworthy for research than a general chatbot: it will not blend in outside knowledge unless you explicitly ask it to.

You can select or deselect individual sources before asking a question, which narrows the answer to just those documents. That's useful when a notebook holds ten sources but your question only concerns three of them.

• Answers are grounded only in your added sources, with citations
• Click any citation to jump to the original passage
• Toggle individual sources on/off to narrow which ones NotebookLM reads

How Do You Generate an Audio Overview in NotebookLM?

You generate an audio overview by opening the Studio panel and clicking Generate under Audio Overview, which produces a conversational two-host discussion of your sources.

NotebookLM turns your documents into a podcast-style back-and-forth, with two AI voices explaining, questioning, and summarizing the material like two people talking through it. It's built for listening on a commute or while doing something else, not for citing as a research source itself.

You can steer the audio overview before generating it. A prompt field lets you tell it to focus on a specific chapter, take a specific angle (like explaining to a beginner), or keep it short. Once generated, you can also enable Interactive Mode on newer versions to ask the AI hosts a live follow-up question mid-playback.

• Open Studio panel > Audio Overview > Generate
• Add a custom prompt to focus or shorten the discussion
• Free tier: roughly 3 audio overview generations per day
• Download the finished audio file to share outside NotebookLM

How Do You Share a NotebookLM Notebook?

You share a notebook by clicking the Share button and choosing between private, shared with specific people, or public link access.

A closed padlock icon marks a private notebook only you can see. Sharing with specific people switches that icon to a shared-person icon, and those collaborators can view sources, chat, and generated content depending on the permission level you set. A globe icon marks a public link, which anyone with the URL can open.

For business or client work, keep notebooks private or shared with named people only. A public link exposes every source document inside the notebook, not just your summary, so treat it the same way you would treat sharing the raw files themselves.

• Private (lock icon): only you can access
• Shared with people (person icon): named collaborators can view/chat
• Public link (globe icon): anyone with the URL can open it, including sources

What Mistakes Do People Make When Using NotebookLM?

The most common NotebookLM mistake is dumping unrelated documents into one notebook, which dilutes answer quality even though nothing breaks visibly.

NotebookLM does not fail loudly when a notebook is too broad. It will still answer your question, just with less precision, because it has to reconcile more competing context. Splitting by project instead of jamming everything into one "main" notebook fixes this without any settings change.

The second common mistake is treating a public share link as harmless. Because the link exposes underlying sources, not just an AI-generated summary, sharing a notebook built on a client's internal documents can leak the documents themselves.

• Mixing unrelated topics in one notebook weakens answer quality
• Public links expose the raw sources, not just the summary
• Forgetting a Drive source lost access still counts toward your 50-source cap

When Should You Use NotebookLM Instead of a General Chatbot?

Use NotebookLM when you need answers grounded in a specific, fixed set of documents rather than general knowledge, and use a general chatbot like ChatGPT or Gemini when your question is open-ended.

NotebookLM's strength is refusing to make things up beyond your sources. That's exactly what you want when studying for an exam from assigned readings, prepping talking points from a client's actual contract, or summarizing a stack of research papers. A general chatbot will happily answer from its training data too, which is faster for broad questions but riskier when accuracy against one specific document matters.

When we evaluate AI tools for a client's workflow at Layer3Labs, the deciding question is usually whether the task needs a fixed, auditable source set or open-ended reasoning. NotebookLM wins the former; a general-purpose model wins the latter. Most teams end up using both, not choosing one permanently.

• NotebookLM: research grounded in a fixed document set, with citations
• General chatbot: open-ended questions, brainstorming, no fixed source list
• Many workflows use both, switching tools by task