Google’s source-grounded research and knowledge tool, renamed Gemini Notebook in July 2026. Unlike general chatbots, it answers only from the sources you upload—PDFs, web pages, YouTube videos, audio recordings, EPUBs—and explicitly tells you when your sources don’t contain the answer, making it unusually trustworthy for research where citations matter. The 2026 update introduced Gemini 3.5 and Antigravity integration, giving every notebook a secure cloud Linux VM that can write and run code for data analysis, charting, and report generation. You can now start with a question instead of pre-gathering sources, with the system suggesting relevant material via Google Search. Standout features include Audio Overviews (podcast-style discussions), Video Overviews (Cinematic/Explainer/Brief, powered by Veo 3 and Nano Banana Pro), slide decks with PPTX export, and interactive flashcards/quizzes with saved progress. Free tier: 100 notebooks, 50 sources each, 50 chats/day, 3 audio/video generations/day. Paid access is bundled into Google AI plans (Plus from ~$4.99/mo, Pro, Ultra). Workspace and Enterprise plans exclude your data from model training.

Pros & Cons:

Pros: The source-grounding discipline is the core differentiator—it refuses to answer beyond your materials rather than hallucinating, which is rare among AI tools and makes it suitable for academic, legal, and professional research where every claim must be traceable; Audio and Video Overviews are genuinely useful learning aids, with the Cinematic video option producing structured, visually rich explanations from your sources; the 2026 addition of a cloud Linux VM means it can now execute code to analyze spreadsheets, generate charts, and produce downloadable reports, extending it from “reading assistant” to “analysis executor”; the free tier is generous enough for individual daily use (100 notebooks, 50 chats/day), and Workspace/Enterprise plans provide strong data-privacy guarantees; Gemini 3.5 + Antigravity integration improved research quality, with Google’s internal benchmark showing 78.2% win rate over the previous version on advanced web research tasks.

Cons: The product’s identity is now broader and harder to pin down—it started as a note/research tool, and now includes code execution, video generation, and agentic research, which may overwhelm users seeking a simple “chat with my PDFs” experience; the most advanced features (code execution, Cinematic Video, voice conversation) are gated to Google AI Ultra subscribers or Workspace customers with AI Ultra Access, so most free users won’t see them immediately; it is Google-ecosystem-bound—there is no self-hosted option, and data lives on Google Cloud, which may be a blocker for organizations with strict data-residency requirements; source limits are real—50 sources per notebook on Free, and a 500K-word/200MB per-source cap, which constrains large document collections; independent third-party validation of the post-Gemini-3.5 version is still accumulating, and some features (voice conversation, mobile recording) were rolled out unevenly by region and plan.

Who is it for?:Students and researchers who need to study, summarize, and interrogate dense documents (papers, textbooks, lecture recordings) with citations; professionals analyzing large PDFs, reports, or meeting transcripts who need answers grounded strictly in those materials; podcasters and content creators who want to turn research into Audio or Video Overviews for publishing; teams already in the Google Workspace ecosystem that need a shared research tool with enterprise data protections; users who want to start a research project from a question rather than building a source library first. Not suited for users seeking a general-purpose web search engine (it deliberately doesn’t search beyond your sources), teams requiring self-hosted or non-Google infrastructure, or free-tier users who need the advanced code-execution and video-generation features.