NotebookLM

Google’s free AI research assistant. Ingest PDFs, YouTube videos, Google Docs / Slides, audio files, and pasted text as sources for a notebook → ask grounded questions, generate podcast-style audio overviews, build mind maps, timelines, and FAQ syntheses. Powered by Gemini 2.5 Flash (the thinking variant) since the May 2025 update.

  • Vendor: Google
  • Site: notebooklm.google.com
  • Model: Gemini 2.5 Flash
  • Pricing: free; 100 notebooks, 50 sources/notebook, 50 chat queries/day, no credit card
  • Mobile: iOS + Android (launch 2026-05-20 per source)

What it does

CapabilityNotes
Multimodal source ingestPDFs, YouTube, Google Docs/Slides, audio files, pasted text
Grounded chatAnswers cite specific source passages
Audio overviewsPodcast-style summaries in 50+ languages / 180+ regions
Mind mapsVisual structure of source content
TimelinesChronological view of source events
FAQ synthesisAuto-generated Q&A from source corpus

Why it matters

NotebookLM is the wiki’s canonical “consumer CAG” product: you upload a bounded corpus, Gemini reads the whole thing per query (no chunking, no vector retrieval). It’s CAG packaged for non-developers. The audio overview feature is also the first wiki entry where podcast is the output format — connects loosely to DeerFlow’s research-output thread.

Sibling product to Google Jules and Google Stitch in Google’s coordinated free-Gemini push. The three together span code, design, and research — Google’s full-stack play against entrenched paid tools (Codex/Copilot, v0/Figma, ChatPDF/Mendeley).

Caveats

  • Free tier limits are real: 50 chat queries/day will hit fast for serious research. Worth tracking whether the NotebookLM Plus tier ships at a sensible price.
  • Gemini 2.5 Flash, not Pro: the thinking variant is fast but not the strongest Gemini model. For very complex synthesis, dump the same corpus into a Gemini 2.5 Pro session via AI Studio instead.
  • Bounded corpus only: not a general web research tool. Bring your own sources.

Sources

See Also