NotebookLM tutorial — your first notebook
Jumping-In Points
This tutorial runs 10 numbered steps from sign-in to finished audio overview. Each step lists exactly what to click and what result to expect. The whole sequence takes under 15 minutes with a single PDF source ready to upload.
The AI research notebook is built around a simple loop: upload material, ask questions, save what matters, share the results. This tutorial walks that loop in full for a first-time user. No prior experience with AI tools is assumed. By the end of step ten you will have a working notebook, at least one saved note, and a generated audio overview ready to play or download.
Before you begin
Gather one or two PDFs or a public web URL on a subject you know reasonably well. Familiarity with the source material lets you judge whether the citation tool is citing accurately. Have a Google account ready. The product page loads in any modern browser — no extension, app, or plugin is required for this walkthrough.
The 10 steps
Step 1 — Sign in
Open the product home page and click "Sign in with Google." Choose any personal or Workspace Google account. The home screen shows your existing notebooks (empty on a fresh account) and a prominent "New notebook" button.
Step 2 — Create a new notebook
Click "New notebook." A blank workspace opens. The layout has three zones: a source panel on the left, a chat pane in the centre, and a notes panel on the right. At this point all three zones are empty.
Step 3 — Name the notebook
Click the default title at the top of the page and type a descriptive name — something like "Tutorial run — April 2026." Naming takes five seconds but pays dividends once you have a dozen notebooks. Press Enter to confirm.
Step 4 — Add your first source
In the left panel, click "Add source." A dialog offers several input types: Upload file, Google Drive, Paste text, and Link (URL or YouTube). Select "Upload file" and choose the PDF you prepared. The source appears in the panel with a pulsing status indicator while it indexes.
Step 5 — Watch indexing complete
Indexing a typical PDF takes 15–30 seconds. When the status indicator clears, the source is fully embedded and ready for retrieval. A small word-count badge appears next to the source title. If you uploaded a very large document, indexing may take up to two minutes.
Step 6 — Ask your first question
In the chat pane, type: "Summarise the main argument of this document in three bullet points." Press Enter. Within a few seconds the assistant returns a bulleted response. Each bullet carries a citation marker — a small superscript number. Click any number to jump to the exact passage in the source panel where that claim originates. This is the grounding mechanism that makes the tool trustworthy for research use.
Step 7 — Ask a follow-up question
Continue the conversation: "What evidence does the author provide for the second point?" The assistant draws on the same indexed corpus and cites the relevant section. Notice that it does not invent evidence — if the document does not address a point, it says so rather than fabricating an answer.
Step 8 — Pin a response to notes
Hover over any chat response and click the pin icon that appears in the top-right corner of the response card. The response is immediately copied to the notes panel on the right. You can edit the pinned note directly: click into the text, make any changes, and the edits save automatically.
Step 9 — Generate an audio overview
At the top of the left panel, find the "Generate" button. Click it and select "Audio overview." The model processes your source corpus and, within two to four minutes, returns a conversational two-host dialogue that summarises the material. Click play to listen in the browser, or click the download icon to save the MP3. This feature is available on the free tier with a daily generation cap. The overview typically runs 8–14 minutes for a single-source notebook. See the Stanford AI portal for broader context on how generative audio tools are evaluated in research settings.
Step 10 — Export your notes
In the notes panel, click the three-dot menu and select "Export to Google Docs." A new Doc opens in your Drive containing all your pinned notes in order. This is the standard handoff path for anyone who needs to share the output with a supervisor, editor, or colleague who does not have access to the notebook itself.
| Step | Click / action | Expected result |
|---|---|---|
| 1 — Sign in | Sign in with Google | Home screen with New notebook button |
| 2 — Create notebook | New notebook | Blank 3-panel workspace |
| 3 — Name | Click title, type name | Custom title saved |
| 4 — Upload source | Add source › Upload file | Source appears in left panel |
| 5 — Indexing | Wait for status clear | Word-count badge appears |
| 6 — First question | Type prompt, Enter | Bulleted answer with citations |
| 7 — Follow-up | Continue conversation | Source-cited follow-up answer |
| 8 — Pin note | Hover response, click pin | Response copied to notes panel |
| 9 — Audio overview | Generate › Audio overview | Playable MP3 dialogue |
| 10 — Export | Notes menu › Export to Google Docs | New Google Doc in Drive |
What to try next
Once you are comfortable with the single-source flow, add a second source and ask the assistant to compare the two documents. Then try the briefing document and study guide templates from the Generate menu. Each template applies a different prompt structure to the same corpus, so the outputs are complementary rather than redundant.
Leander C. Tveitmoen, Consulting Analyst at Stonegate Research Partners in Trondheim, uses the tutorial flow as a 15-minute onboarding exercise for new analysts: "We run every new researcher through steps one to ten on their first afternoon. By step six they understand the citation model. After that the tool sells itself."
Tutorial — frequently asked questions
Short answers to what new users ask after running through the steps for the first time.
Do I need a paid account to follow this tutorial?
No. Every step in this tutorial works on the free tier. A Google account is the only requirement. The paid tier raises usage caps but is not needed to complete the walkthrough.
How long does indexing take after I upload a source?
Most sources index within 10–40 seconds. Longer PDFs or documents with complex formatting may take up to two minutes. A status indicator next to each source name clears when indexing is complete and the source is ready to query.
Can I add more sources after I have already started chatting?
Yes. Sources can be added, removed, or temporarily hidden at any time. The model re-indexes the updated corpus automatically and existing notes and chat history are preserved throughout.
What file types work best for a first tutorial?
A PDF or a Google Doc with clear prose works best for a first run because the citation previews are straightforward to verify. YouTube links and public URLs also work well and require no file download on your end.
Want the full picture before you start?
The complete guide covers every stage of a notebook lifecycle — from setup and source management through advanced chaining techniques.
Read the complete guideWhere this tutorial fits in the site
The tutorial is the fastest path from zero to a working notebook. For a more narrative introduction, start with the NotebookLM guide. To see the tool applied to a real research scenario before you try it yourself, the NotebookLM demo walks a three-source climate-research project from upload to output. Users curious about what is happening under the hood while they follow these steps should read the AI primer on the retrieval-augmented generation loop. For pricing context before committing to a workflow, see the pricing page and the Plus tier detail.
After running this tutorial on a test corpus, most users want to try more source types. The sources and uploads guide covers every accepted format in detail. The in-depth review offers an honest assessment of where the tool excels and where it has gaps so you can decide whether it fits your specific workflow before investing significant time uploading a large corpus.