The short answer (July 2026): NotebookLM is Google's source-grounded AI research tool: you upload your own documents, PDFs, YouTube videos, and other sources, then ask questions and generate outputs -- Audio Overviews, Video Overviews, mind maps, flashcards, and slide decks -- that draw only from what you gave it. It runs at notebooklm.google and is free to start with a Google account; paid tiers unlock higher limits through Google AI subscriptions starting at $4.99/month. The core value proposition is that responses are grounded in your sources and cited, not drawn from the open web or general training data.
Last verified: July 14, 2026, against official docs and pricing pages.
What NotebookLM Actually Does
NotebookLM organizes your research into "notebooks," each one a closed knowledge base built from sources you add. When you ask a question, the model searches and synthesizes only that notebook's content, then cites the exact passage it drew from. This is meaningfully different from a general-purpose chatbot: you control the knowledge base, and hallucinations about facts outside your sources are structurally harder to produce.
The Studio panel on the right side of each notebook is where you generate outputs. It is separate from the chat panel on the left, where you ask questions. You can run both simultaneously -- listening to an Audio Overview while browsing a Mind Map, for example.
NotebookLM does not use your uploaded content to train Google's models, according to Google's official statement on the Workspace product page.
Step 1: Create Your First Notebook
Go to notebooklm.google.com and sign in with any Google account. Click "New notebook" in the top-left corner. You get a blank Sources panel on the left, a chat interface in the center, and a Studio panel on the right.
Name the notebook immediately. Titles are searchable and you will accumulate notebooks fast.
The free tier supports 100 notebooks with up to 50 sources each, per Google's official FAQ.
Step 2: Add Sources
Click the "+" button in the Sources panel. NotebookLM accepts a wide range of formats, confirmed in Google's official help article on adding sources:
- PDFs, Word documents (.docx), PowerPoint files (.pptx)
- Google Docs, Google Slides, Google Sheets (auto-sync every few minutes)
- Audio files (MP3, WAV) and YouTube video URLs (requires public video with captions)
- Web URLs, images (JPEG, PNG, WebP), CSVs, EPUB files
- Plain text (copy/paste) and Gemini Chat exports
Each source can contain up to 500,000 words or 200MB, whichever comes first. Google Sheets are capped at 100,000 tokens; Google Slides at 100 slides. Copy-protected PDFs will fail to import.
One practical tip: add sources in batches. NotebookLM processes them in parallel, so uploading 10 sources at once is faster than uploading them one by one.
Step 3: Chat With Your Sources
Once sources are loaded, type questions in the center chat panel. NotebookLM returns answers with inline citation markers -- click any marker to see the exact passage in the source it came from. This citation system is the fastest way to verify the model's claims before using them downstream.
You can ask it to compare arguments across multiple sources, identify contradictions, extract all mentions of a specific term, or draft a summary in a particular voice. The model stays within the sources you gave it; if you ask about something not covered in your documents, it will say so rather than improvise.
Chat history is saved automatically across sessions. You can resume previous conversations from the same notebook.
The free tier allows 50 chat queries per day.
Step 4: Generate Studio Outputs
The Studio panel produces six distinct output types. Select any tile to start generation:
Audio Overviews: A synthesized podcast-style conversation between two AI hosts covering your sources. Options include four formats (Deep Dive, The Brief, The Critique, The Debate), length controls (Shorter / Default / Longer), 80+ languages, and a custom prompt field to steer focus or expertise level. Interactive mode lets you ask follow-up questions mid-listen, though this is currently English-only. Per Google's Audio Overview help page, voices are AI-generated and may contain inaccuracies.
Video Overviews: Narrated slide presentations generated from your sources, with customizable visual style (whiteboard, kawaii, watercolor, classic), format (cinematic, explainer, or short), and language. Explainer format available in 80+ languages; Cinematic and Short are English-only.
Mind Maps: Visual diagrams showing topic connections and subtopics extracted from your sources. Exportable as PNG.
Reports: Tailored documents -- briefing docs, study guides, blog posts, slide decks. Slide decks export as PPTX.
Flashcards and Quizzes: Spaced repetition cards and multiple-choice quizzes generated from source content. Progress saves across sessions; you can mark cards as mastered or flag them to review again.
You can store multiple outputs of the same type within one notebook -- for example, Audio Overviews in different languages for different audiences.
Step 5: Use Deep Research
Deep Research is a separate mode accessed by clicking "Web" in the Sources panel. Instead of you finding sources, NotebookLM browses hundreds of websites on your behalf, creates a structured report with citations, then lets you import the report and all cited sources into your notebook with one click.
The feature was added in November 2025, per Google's announcement. It runs in the background, so you can continue chatting or adding other sources while it works. The resulting report becomes another source you can then convert into an Audio Overview, Mind Map, or any other Studio output.
Free users get 10 Deep Research queries per month. Plus users get 3 per day; Pro users get 20 per day.
Step 6: Share Notebooks
To share a notebook, click "Share" in the top-right corner. You can set access to "Anyone with the link" for public sharing, or invite specific people by email. Recipients can be assigned either Viewer (read, chat with sources, but no editing) or Editor (full access including adding sources and generating outputs) roles.
Public sharing is available on all tiers. Notebook sharing as a feature appears across plans, though collaboration for enterprise teams routes through Google Workspace or NotebookLM Enterprise.
Plan Comparison
NotebookLM is not sold as a standalone subscription. Paid tiers are bundled with Google AI subscriptions. All prices are monthly (US); annual billing may differ.
| Plan | Monthly price | Notebooks | Sources/notebook | Chat queries/day | Audio Overviews/day | Deep Research/day |
|---|---|---|---|---|---|---|
| Standard (Free) | $0 | 100 | 50 | 50 | 3 | 10/month |
| Google AI Plus | $4.99 | 200 | 100 | 200 | 6 | 3/day |
| Google AI Pro | $19.99 | 500 | 300 | 500 | 20 | 20/day |
| Google AI Ultra (20 TB) | $99.99 | 500 | 500 | 2,500 | 100 | 75/day |
| Google AI Ultra (30 TB) | $200 | 500 | 600 | 5,000 | 200 | 200/day |
Source: Google's upgrade page and Google One plans page. Limits are confirmed for the Standard (Free) tier; paid tier specifics are sourced from Google's support documentation and may update.
The Plus tier is the right first upgrade for most users: 200 daily chats and 100 sources per notebook covers the majority of real research workflows. Pro makes sense if you are running Deep Research sessions daily or managing large document libraries.
Practical Workflows That Work Well
Contract or policy review: Load the contract PDF plus any relevant regulatory docs. Chat to locate specific clauses, generate a briefing doc that flags obligations, then export as PPTX for a stakeholder presentation -- all grounded in the actual documents.
Learning a new technical domain: Add a handful of authoritative papers or documentation pages. Generate an Audio Overview in "Deep Dive" format for background listening, then switch to flashcards to test recall. Add new sources as you go deeper; the notebook grows with you.
Podcast or meeting prep: Drop in a guest's published work, relevant articles, and an agenda doc. Generate an Audio Overview using "The Critique" format to surface counterarguments, then use chat to draft specific questions.
Research synthesis: Run Deep Research on your core question, import the report and its sources, then generate a mind map to visualize how the sources connect. Export the mind map as PNG for a presentation.
See also: how to build AI agents if you want to go beyond document Q&A toward automated pipelines.
FAQ
Can NotebookLM access the internet or sources outside my notebook?
Only when you explicitly use the Deep Research feature to pull web sources. In standard chat mode, NotebookLM draws exclusively from the sources you have added to that notebook and will tell you when a question falls outside them.
Does NotebookLM work for audio and video content?
Yes. You can upload MP3 and WAV audio files directly, and add YouTube URLs as sources (the video must be public and have captions). NotebookLM processes the transcript or audio content. There is no video length limit unless the caption file exceeds 500,000 words.
What happens to my sources if I hit the limit?
You can delete sources to stay under the limit. Free users cap at 50 sources per notebook; uploaded files are not automatically purged. Deleted sources -- or notes -- cannot be recovered per Google's FAQ.
Is NotebookLM available for schools and enterprise?
Yes. Google for Education offers NotebookLM free for qualifying institutions. Enterprises can provision NotebookLM through Google Cloud's NotebookLM Enterprise or qualifying Google Workspace plans, with additional data protection and admin controls.
Can I use NotebookLM on mobile?
Yes, the mobile app is available for iOS and Android through standard app stores. Some Studio features are more limited on mobile compared to the desktop web version.
If this guide was useful, get AI Weekly free, 3 issues a week, read by 40,000+ practitioners.