How to Use NotebookLM (Now Gemini Notebook) to Summarize YouTube Videos and PDFs

Starter · Level 1 of 3

You already have a free tool that can watch a long YouTube video or read a dense PDF and hand you a clean summary you can question further: Google’s NotebookLM. In mid-2026 Google renamed the product to Gemini Notebook, but the tool, the free tier, and the notebooklm.google.com address are unchanged. Paste in a YouTube link or upload a PDF, and it treats that source as the only ground truth, summarizes it, and answers questions strictly from what you gave it, at no cost and with no paid summarizer subscription required.

What Gemini Notebook (Formerly NotebookLM) Actually Does

Gemini Notebook is a source-grounded notebook, not a general search engine. You give it specific sources — PDFs, web pages, YouTube videos, Google Docs, audio files — and it answers questions using only those sources, plus two optional web-search features covered further down. That distinction matters: ask it something outside the sources you added, and it tells you the sources don’t cover that, rather than guessing from general knowledge the way a general-purpose chatbot would.

This is what makes it a solid stand-in for a paid summarizer. Instead of paying for a tool built specifically to summarize videos or documents, you get the same job done inside a notebook that also lets you ask follow-up questions, cross-reference multiple sources, and keep a running research thread.

What You Need Before You Start

  • A free Google account. Any Gmail address works — no separate NotebookLM or Gemini Notebook sign-up.
  • A public YouTube video with captions, or a PDF (Word, plain text, Markdown, CSV, PowerPoint, and ePub files work the same way).
  • Availability in a supported country. Gemini Notebook is available in-browser to users above the age of consent in Google’s list of supported countries and territories; the mobile app supports a narrower list.
  • Awareness of the free-tier caps: 50 sources per notebook, each source up to 500,000 words or 200MB, and 50 chat queries a day.

The source list is broader than most people expect: Google Docs, Google Slides (up to 100 slides), Google Sheets (currently capped around 100,000 tokens per sheet), Word, plain text, Markdown, CSV, PowerPoint, and ePub files, pasted text, audio files like MP3 and WAV, images in common formats, and web URLs. Web pages only import the visible text, not embedded videos or paywalled content, and a PDF linked through a URL is treated the same as an uploaded PDF.

Step 1: Add a YouTube Video as a Source

  1. Go to notebooklm.google.com and sign in with your Google account.
  2. Select “New notebook,” or open an existing one if you’re adding to a running research thread.
  3. Click “Add” in the Sources panel.
  4. Choose the YouTube source type from the list, then paste the video’s public URL.
  5. Wait for the import. Gemini Notebook pulls only the video’s text transcript, not the audio or video itself, so processing is usually quick even for a long video.

A few things will trip you up here. Only public videos with captions, either creator-uploaded or YouTube’s auto-generated ones, will import. A video uploaded in the last 72 hours may not be available yet. A video with no speech, a silent screen recording, for example, won’t import at all. And if the original video is later deleted or set to private, Gemini Notebook removes it from your notebook automatically within 30 days.

Step 2: Add a PDF as a Source

  1. In the same Sources panel, choose the option to upload a file, or drag the PDF straight into the panel.
  2. Select the PDF from your device. It can be up to 200MB or 500,000 words, whichever limit you hit first.
  3. Gemini Notebook extracts the text and adds it as a source you can view side by side with your chat.

The same panel accepts Word documents, plain text, Markdown, CSV, PowerPoint, and ePub files the same way, so a single notebook can mix a YouTube lecture, a PDF whitepaper, and your own notes without any extra setup.

Getting a Summary and Asking Follow-Up Questions

Open any source and Gemini Notebook shows an auto-generated summary in the Source Guide panel, no prompting required. For a tighter summary, ask directly in chat rather than relying only on the auto-summary: “What are the three main arguments in this video?” works better than a bare “summarize this.”

Once you have more than one source, mention the source by name in your question to keep Gemini Notebook from blending answers across everything in the notebook. “According to the PDF, what does the author recommend for beginners?” narrows the answer to that one document instead of pulling from the YouTube video too.

A concrete version of this: load a 40-minute conference talk as a YouTube source and a 20-page PDF whitepaper on the same topic into one notebook. Ask “Where do the talk and the whitepaper disagree?” and Gemini Notebook checks both sources for you and answers with the specific claim from each, rather than you scrubbing through a video timeline and a PDF side by side to catch the difference yourself.

Beyond chat, the Studio panel can turn the same sources into a mind map, a set of flashcards, or a short audio overview modeled as a two-host discussion of your material, useful if you’d rather listen to a synthesized rundown on a walk than read one, though these all draw from the same daily quotas as the free tier’s other limits.

Where Gemini Notebook Hits a Wall

The numbers below come from Google’s Gemini Notebook Help documentation, not a cached blog post, because free-tier limits on AI tools change often enough that a specific figure is worth sourcing directly rather than repeating from memory.

Limit (free Standard tier)Current capSource
Sources per notebook50Google — Upgrade Gemini Notebook
Chat messages50 / dayGoogle — Upgrade Gemini Notebook
Audio overviews3 / dayGoogle — Upgrade Gemini Notebook
Video overviews3 / dayGoogle — Upgrade Gemini Notebook
Words or size per source500,000 words / 200MBGoogle — Add or discover new sources

Accessed 2026-07-27. Google describes these usage limits as subject to change, so treat the exact numbers as a snapshot rather than a permanent ceiling.

If you’re summarizing one video or PDF at a time, you won’t come close to those limits. If you’re trying to build a large research library inside one notebook, you will.

YouTube import has its own failure modes worth knowing before you rely on it: no captions file means no import, a dubbed-only audio track is not a substitute for a captions file, and a video without any speech won’t process. Import can also fail on an invalid link or a video flagged as unsafe.

Treat the summaries themselves as a strong starting point, not a final answer, the same caution that applies to any AI-generated output. If a source document is messy, poorly scanned, or has inconsistent formatting, questions and answers about it can drift or miscite. For anything you’re relying on to make a decision, spot-check the answer against the actual source before you act on it.

Gemini Notebook also isn’t a general search engine by default. The two features that do reach outside your added sources, Fast Research and Deep Research, work by searching the web or your Google Drive and let you choose what to import; they don’t silently blend outside information into answers about your existing sources.

Privacy Considerations for What You Upload

For a personal Google account, Google states that content in Gemini Notebook is not used to directly train its foundational AI models unless you submit thumbs-up or thumbs-down feedback on a response. If you do submit feedback, the associated content, your prompts, sources, uploads, and the output, is reviewed by specially trained teams to address the issue, but your feedback is disconnected from your Google Account before reviewers see it, and reviewed feedback data is retained for up to three years under that disconnected state, per Google’s Gemini Notebook privacy documentation.

If you use Gemini Notebook through a qualifying Google Workspace or Workspace for Education account, the protection is stronger: your uploads, queries, and responses are not reviewed by human reviewers at all, even when you submit feedback, and are not used to train models.

The practical rule either way: don’t upload a document you don’t have the rights to, and treat anything you paste in as visible to Google’s systems, even on the free tier where reviewed training use only kicks in if you actively submit feedback.

When to Use This vs. a Paid Summarizer, and When to Skip It

Use free Gemini Notebook when you have a handful of videos or PDFs and want one place to summarize them and keep asking questions afterward. That single workflow, upload, summarize, then interrogate, is exactly what most people are paying a dedicated summarizer app to do.

Consider Google AI Plus, currently priced at $4.99 a month, when the 50-source cap or 50-chats-a-day limit actually gets in your way, since it raises the per-notebook source cap to 100 and the daily chat allowance to 200. Google’s AI plan pricing and limits have shifted more than once this year, so check the current numbers on Google’s own pricing page before you upgrade.

Skip Gemini Notebook altogether if you need a tool built for team-wide document management at scale, or a specialized feature it doesn’t offer, like multi-speaker transcription labeling for long recordings. For one person trying to get through YouTube videos and PDFs without paying anything, it’s hard to justify a separate subscription first.

If you already lean on Google Gemini for everyday tasks, Gemini Notebook fits into the same account and habit rather than adding a new login to manage. And if you’d rather standardize a single AI tool for organizing ongoing work instead of research notebooks, see how ChatGPT Projects keeps work organized in one place for a different approach to the same underlying problem: too many scattered documents and no single place to ask questions about them.

None of this requires deciding on a tool forever. A notebook costs nothing to create, so the lowest-risk way to judge whether Gemini Notebook replaces a paid summarizer for you is to load in the next long video or PDF you were about to skim and see whether the follow-up questions actually save you time.