NotebookLM's value depends entirely on getting useful sources in. Out of the box, NotebookLM supports a limited set of source types. With the right tools, you can save virtually anything from the web into a NotebookLM notebook. Here's the complete guide.
Native NotebookLM Source Types
You can add these directly through the NotebookLM Add Sources interface:
- **PDFs** (up to 200 MB / 500K words)
- **Plain text files** (.txt)
- **Markdown files** (.md)
- **Google Docs** (linked from your Drive)
- **Google Slides** (linked from your Drive)
- **Web URLs** (NotebookLM fetches and extracts text)
- **YouTube URLs** (transcript automatically extracted)
- **Audio files** (auto-transcribed)
- **Pasted text** (any length up to 500K words)
That's the native surface. For everything else, you need workarounds.
AI Chat Conversations
Native: paste into a text source.
Better: [Notebook Toolkit](/download) one-click capture from:
- ChatGPT
- Claude
- Gemini
- Perplexity
- DeepSeek
- Mistral / Le Chat
- Microsoft Copilot
- Grok
- Pi
- HuggingChat
- Poe
- Phind
- You.com
- Meta AI
The capture preserves formatting, code blocks, citations (in Perplexity case), and conversation structure.
Social Media
Native: copy-paste into a text source (loses structure).
Better: Notebook Toolkit captures:
- **Reddit threads** (with nested comments)
- **Twitter / X tweets and threads** (with quote tweets)
- **LinkedIn posts and articles**
- **Hacker News stories and comments**
Each capture preserves the author, timestamps, and reply structure.
Video Content
Native: paste YouTube URL.
Better: Notebook Toolkit captures YouTube with full transcript + timestamps + chapter info.
For non-YouTube video (Vimeo, podcasts, etc.):
- Use a transcription service (Otter, Notta, MacWhisper for local)
- Save transcript as PDF or text
- Add to NotebookLM
For TikTok and short-form video: transcribe the audio, paste the transcript.
Articles & Blogs
Native: paste URL.
Better: Notebook Toolkit captures with reader-mode extraction (strips ads/nav, preserves formatting), including:
- Medium articles
- Substack newsletters
- WordPress / Ghost blogs
- Any web page
Academic Sources
Native: download PDF from arXiv / journal site, upload.
Better: Notebook Toolkit captures from arXiv and PubMed with metadata (authors, abstract, DOI, BibTeX) directly into NotebookLM.
For Google Scholar or institutional databases: download PDFs, upload natively.
Wikipedia & References
Native: paste URL.
Better: Notebook Toolkit captures the article with section structure and citation list preserved.
Native: copy/paste into text source.
Better: forward emails to a Gmail-monitored address, save as PDF, upload. Or use Gmail's "print to PDF" for individual emails.
For email digests (Morning Brew, Stratechery, etc.): subscribe to web versions, capture via Notebook Toolkit.
Code Repositories
Native: paste a README.
Better: Notebook Toolkit captures GitHub READMEs, issues, discussions, and gists.
For deeper code analysis: clone the repo, ask Claude/ChatGPT to summarize, save the summary as a source.
Documents (Word, PowerPoint)
Native: convert to PDF first, then upload.
Workarounds:
- Word .docx → File > Save As > PDF, then upload
- PowerPoint .pptx → File > Export > PDF, then upload
- Or upload to Google Drive as Docs/Slides, then link
ePub is not supported natively; convert via Calibre to PDF or text.
Audio / Podcasts
Native: upload audio files (auto-transcribed).
Workflow:
- Download podcast MP3 from Spotify (where allowed) or use a podcast-app's export
- Or transcribe with Otter / Whisper, upload transcript
- For radio shows / interviews: same approach
Long podcasts (3+ hours): split into chapters if hitting size limits.
Tweets / X Threads
Notebook Toolkit captures threads with quote tweets, replies, and media references. Without Notebook Toolkit, you can:
- Use a thread-unroller (e.g., threadreaderapp.com)
- Copy the unrolled thread, paste as text source
Slack / Discord / Forum Threads
Native: copy-paste relevant messages into a text source.
Workflow: most threads with valuable info should be summarized first. Use ChatGPT to summarize a Slack thread, then save the summary to NotebookLM.
For Discord: take screenshots → OCR → paste text. Or paste manually.
News Articles Behind Paywalls
If you have legitimate access:
- Save as PDF via browser print-to-PDF
- Use Notebook Toolkit (works with most paywalled sites if you're logged in)
- Upload PDF
For research access (academic journals, paywalled trade publications), institutional access is your friend.
Books
Native: PDF if you have one.
For ePub: convert via Calibre.
For Kindle: export highlights via the Kindle Notebook page (read.amazon.com/notebook), copy to text source.
For physical books: type or dictate key passages into a text source. Or scan and OCR.
Spreadsheets & Data
Native: convert to CSV or text, paste.
For tables: Markdown tables work well as text sources.
For complex data: summarize key findings in a doc, upload that.
Images
Native: not supported as image sources.
Workaround: extract text via OCR (Google Vision, Adobe, or built-in macOS/iOS OCR), paste text.
For research figures: caption + your annotation works well.
Best Practices for Sources
1. Quality over quantity: . 30 great sources beat 300 mediocre ones.
2. Curate before uploading: . Skim the source. Is this really relevant? If not, skip.
3. Tag in notes: . NotebookLM doesn't have native tags. Write a note inside the notebook describing source clusters ("Foundational: 1-5; Recent extensions: 6-12").
4. Reload stale sources: . Web URLs go stale. Refresh important ones periodically.
5. Mix media types: . The best notebooks combine PDFs, articles, AI chats, and videos. Each medium adds different angles.
When to Use Notebook Toolkit
If you regularly capture from:
- More than one AI chat platform
- YouTube + articles + social
- Academic sources
...Notebook Toolkit pays for itself in saved time within the first week.
The free tier of Notebook Toolkit covers most workflows. The Pro tier ($5/month) is worth it once you're capturing 20+ sources per week.
Bottom Line
NotebookLM accepts almost anything you put in front of it — if you can get the content into a supported format. Native upload covers the basics; Notebook Toolkit closes the gaps for AI chats, video, social media, academic sources, and more.
A high-leverage habit: install Notebook Toolkit this week. Capture sources actively for two weeks. By the end of the month, your NotebookLM library will be 3-5x larger and dramatically more useful.
Source ingestion is the bottleneck. Fix it once, and NotebookLM's value compounds for years.