NotebookLM (formerly "Project Tailwind") is Google Labs' proprietary research assistant designed as a "source-grounded AI" — every answer and generated artifact is restricted to materials the user has personally uploaded to the notebook. Launched in July 2023, over four years it has evolved from an experiment into a rich platform backed by Gemini 3 (replacing Gemini 2.5 Flash in December 2025) with a 1M-token context window and a full multimodal stack.
Key 2026 features: Audio Overviews (podcast-style conversation between two AI hosts about the sources; interactive mode lets you join the conversation), Cinematic Video Overviews (video presentations with narration and visualisation), Deep Research (a multi-hour background agent iterating through sources), Mind Maps, Study Guides, Slide Decks, Infographics, Chat Goals (defining a goal for the session), saved conversation history, 1M-token context. Source support: PDFs, Google Docs, Google Slides, web pages, YouTube (with transcript), audio, pasted text.
Pricing ecosystem (mid-2026): Standard $0/mo (100 notebooks, 50 sources/notebook, 50 chats/day, 3 Audio+Video Overviews/day, 10 Deep Research/month), NotebookLM Plus in Google AI Plus $7.99/mo (200 notebooks, 100 sources, 200 chats, 6 Audio+Video, 3 Deep Research/day), NotebookLM Pro in Google AI Pro $19.99/mo (500 notebooks, 300 sources, 500 chats, 20 Audio+Video, 20 Deep Research/day + Gemini 3), NotebookLM Ultra in Google AI Ultra $99.99/mo (20TB, 500 sources, 2,500 chats, 100 Audio+Video, 75 Deep Research/day, watermark-free) or $200/mo (30TB, 600 sources, 5,000 chats, 200 Audio+Video with 20 Cinematic Video Overviews, 200 Deep Research). Student discount (US, .edu): AI Pro $9.99/mo.
Organisational variants: NotebookLM Plus included as a core service in Google Workspace from Business Standard ($14/user/mo) upward, with enterprise-grade data protection (prompts and uploads are not used for model training). NotebookLM Enterprise sold separately through Google Cloud (~$9/licence/mo, min 15 licences) — adds VPC Service Controls, IAM, audit logging, custom expert guides, organisational notebooks. Consumer user data is also not used for training (Google's declaration) — a key context for academia and regulated sectors.
Competitively, NotebookLM occupies a unique spot in the "grounded research assistant" category — a narrow alternative to general-purpose chatbots (ChatGPT, Claude, Gemini app) and classical document parsers (Elicit, Consensus). Its strength is the fusion of Gemini 3 with the research workflow: multi-format sources → Deep Research → Audio/Video/Slide artifacts. Limitations: no export to CSV/XML (only Google Slides and Docs), no in-notebook custom code, no web crawling beyond individual URLs. Available via notebooklm.google, iOS/Android mobile apps, and Google Workspace domain.
