Most people have tried a general-purpose chatbot. Far fewer have tried an AI assistant that works only from documents they choose. That is the idea behind Google’s NotebookLM, a research tool that reads the sources you upload and answers questions from them, with citations pointing back to the exact passage. In November 2025 Google added a Deep Research mode and broader file support, which widened the range of tasks the tool can handle.
This guide explains how NotebookLM works and walks through eight practical uses.
How it works
A user creates a notebook and adds sources: PDFs, Google Docs, Word files, Slides, Google Sheets, web links, images and YouTube video transcripts. The assistant then answers questions grounded in that material rather than in general internet knowledge. If a notebook contains five papers and two lecture decks, the answers come from those seven sources, and each claim carries a citation showing where it came from. That makes fact-checking the AI far easier than with a standard chatbot.
1. Deep Research for serious projects
Google announced Deep Research in NotebookLM on November 13, 2025. The user types a topic, the tool asks what to focus on, then searches the web for relevant sources and produces a structured report that can be added to the notebook as a source. Google offers two modes: a fast mode for a quick overview and a deep mode for a more thorough investigation. Users can keep adding papers, links and spreadsheets while they work, and the notebook updates to include them.
2. Study guides and FAQs
With a course’s readings loaded, a student can ask for a study guide covering key questions, short answers, definitions and formulas. NotebookLM can also generate flashcards and quizzes from the same material, which can be edited and shared with classmates or a supervisor.
3. Audio Overviews
The feature that first drew wide attention is Audio Overview. Select a few papers or chapters, and NotebookLM produces a conversation between two AI hosts who discuss the material in plain language, in the style of a podcast. The content comes entirely from the user’s own sources, and playback speed is adjustable.
4. Video Overviews and mind maps
The Studio panel can turn uploaded material into a Video Overview, a mind map or a short report. A mind map shows the main ideas at the center with subtopics branching out, which is a fast way to see the shape of an entire syllabus chapter. The output can be exported for use in presentation tools.
5. Shared notebooks for a class, team or client
Google has made it possible to share a notebook by link, in the same way as a Google Doc. Viewers cannot edit the sources, but they can ask the AI questions, request summaries and listen to Audio Overviews. A teacher can build one notebook for a whole class, and a company can load onboarding materials, policy documents and FAQs into a notebook that new employees can query.
6. Working with Google Workspace material
NotebookLM supports Google Drive URLs and Google Sheets as sources, so project proposals, data tables and shared documents can go straight into a notebook. From there a user can ask for the key figures in a dataset, a timeline built from a set of documents or a decision summary drawn from meeting notes.
7. A research hub for content creators
Creators producing educational videos can drop articles, papers and video transcripts into a single notebook and ask for a video outline, short-form scripts or question-and-answer segments. Because every claim is cited, references for a video description are easy to assemble.
8. Exam preparation under pressure
Scattered slides, notes and screenshots can be uploaded as PDFs or images. NotebookLM can then produce chapter-by-chapter summaries, key formulas, likely exam questions and short explanations, and a student can ask follow-up questions in the same chat, all answered from the uploaded notes.
Who should try it
NotebookLM turns a pile of documents into something that can be questioned, summarized and listened to. It suits anyone whose work involves reading and synthesizing material: researchers, students, teachers, analysts and content creators. The free tier is generous enough to test it properly, and higher limits come with Google’s paid AI plans.
