NotebookLM is Google's source-grounded AI research assistant, powered by Gemini, that lets you upload documents, videos, and web content and then chat, summarize, and generate rich outputs β all anchored strictly to your own material rather than the open internet. Its defining feature is RAG (Retrieval-Augmented Generation): every answer cites the exact passage in your sources, slashing hallucination compared to general chatbots. The tool has evolved rapidly from a simple summarizer into a full research studio, adding audio podcasts, video overviews, mind maps, flashcards, slide decks, and agentic Deep Research in 2025β2026. The key mental model is this: NotebookLM knows only what you feed it β garbage in, garbage out β so curating high-quality sources is the single biggest lever for getting great outputs.
What This Cheat Sheet Covers
This topic spans 20 focused tables and 170 indexed concepts. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.
Table 1: Notebook Basics β Creating and Managing Notebooks
Every project in NotebookLM lives inside a notebook β a self-contained workspace with its own sources, chat history, and Studio outputs. Understanding the notebook structure is the foundation for everything else: notebooks are isolated from each other (no cross-notebook queries), and all analysis is limited to the sources you add.
| Feature | Example | Description |
|---|---|---|
Go to notebooklm.google.com β click "Create new notebook" β add sources | β’ Single click to start β’ a summary of all sources is generated automatically in the Chat panel | |
Free: 100 notebooks; Plus: 200; Pro: 500; Ultra: 500 | Each notebook is independent β NotebookLM cannot access information across multiple notebooks at once. | |
Click "+ Add" in the Sources panel, then select file, URL, or Drive | β’ Sources panel shows all imported materials β’ checkboxes let you include or exclude specific sources per query | |
Uncheck "Annual Report Q3" to exclude it from chat | Use per-source checkboxes to narrow context β useful when you want answers only from a subset of documents. |