Google's Expert Intelligence Turns 100,000 E-Books Into AI Study Partners in Gemini Notebook

Google has launched Expert Intelligence, a new feature for Gemini Notebook that turns the e-books you’ve purchased on Google Play Books into queryable AI sources. Announced August 27, 2026, on the official Google blog by Steven Johnson (Editorial Director, Google Labs) and Will Houghteling (Director of Product Management, Expert Intelligence), the feature arrives with more than 100,000 eligible books from major publishers — and a set of publisher-negotiated controls that make it the most serious attempt yet to reconcile AI chat with the book industry’s licensing wall.

The premise is simple: you already own books by people who spent years mastering a subject. Gemini Notebook can now read those books for you — not to replace reading, but to make the expertise inside them computationally accessible. Ask a question, get an answer grounded in the book with citations. Generate a quiz from a chapter, an infographic of its frameworks, or an Audio Overview you can listen to on a commute. Combine one author’s framework with your own journal, your team’s documents, or a photo of the inside of your refrigerator.

What Expert Intelligence actually does

The mechanics work through Gemini Notebook’s existing source system, with an important twist: the e-book becomes a first-class source with licensing attached. From Google Play Books, an eligible title shows a “Tools” badge with Gemini Notebook listed; long-press the cover in your library, choose “About this book,” and select “Use with Gemini Notebook.” The book then flows into your notebook as a grounded source.

Once inside, the book participates in everything Gemini Notebook can do:

• Grounded Q&A — answers with “direct citations to relevant content in her book,” in Google’s example. The model answers from the book, not from its training corpus.
• Artifact generation — Infographics, Audio Overviews (the podcast-style summaries that made NotebookLM famous), Quizzes, and the rest of Gemini Notebook’s artifact toolkit, now generated from a specific author’s text.
• Source mixing — the book sits alongside your PDFs, websites, YouTube videos, audio, and personal notes. Google’s launch examples are pointed: a manager brainstorming feedback strategies with Kim Scott’s Radical Candor; a woman combining The New Menopause by Dr. Mary Claire Haver with her own securely-uploaded journal entries; a parent photographing the fridge and asking for a week of family meal plans based on Michael Pollan’s Food Rules.

That third example is the tell. Expert Intelligence isn’t a chatbot bolted onto a bookstore — it’s RAG (retrieval-augmented generation) applied to your personal library, where the “retrieval” layer is books you paid for and the generation layer already knows how to produce study materials.

The licensing story is the real news

What separates Expert Intelligence from every “upload a PDF and chat with it” tool of the past two years is the deal structure behind it. Google states plainly that it spent “the past several months” meeting with authors and publishers, and the launch list shows who signed: Bloomsbury, De Gruyter Brill, Johns Hopkins University Press, Macmillan Publishers, O’Reilly Media, and Penguin Random House — six of the largest trade and academic houses, contributing over 100,000 books at launch.

The controls negotiated alongside are strict and consumer-side:

1. Ownership requirement. To interact with a book in Gemini Notebook, “we’ve ensured… they must own the book through Google Play Books.” No library loans, no free previews, no “someone shared the file.”
2. Collaborator purchase prompts. If a notebook containing a book is shared, other collaborators are prompted to buy their own copy to engage with the content. Otherwise the book is simply unavailable in their view of the notebook.
3. Discoverability as compensation. Google’s framing is that “over time, Expert Intelligence will create content discovery and help authors reach new and more engaged readers” — the same discovery-loop argument Spotify made to labels, now aimed at publishers.

More than 15 bestselling authors — including Steven Pinker, Michael Pollan, and Jennifer Wallace — also curated Featured Notebooks that supplement their books with additional sources and artifacts, a curated on-ramp showing the engagement Google wants readers to build around a title.

For U.S. users, Google is even giving away a free eligible e-book (promo code EXPERTINTELLIGENCE, while supplies last) to seed the habit.

From NotebookLM to Gemini Notebook: the platform play

Expert Intelligence lands six weeks after Google rebranded NotebookLM to Gemini Notebook on July 16, 2026 — same product, deeper Google integration, plus a secure cloud computer per notebook enabling native code execution. The rename was widely read as a demotion of a beloved standalone brand into the Gemini ecosystem; Expert Intelligence is the first feature that justifies it.

That’s because Expert Intelligence is explicitly “a cross Google initiative that helps you engage with your trusted sources through Google AI products.” Books in Gemini Notebook are only the start: Google says the same licensed-source system will spread to the Gemini app and AI Mode in Search, and that “soon, people will also be able to add more sources like third-party subscriptions, business research reports, and textbooks.”

Read that list again. Third-party subscriptions (news paywalls, journals), business research reports (Gartner, Forrester-class content), and textbooks (the most expensive segment of publishing) are exactly the high-value, high-trust content that general chatbots cannot legally ground on today. If Google signs those categories the way it signed book publishers, AI Mode in Search stops being “Google with a chatbox” and becomes the only assistant with a legal right to cite the sources people actually trust.

Why it matters

For readers and learners, this is the first mainstream grounding of AI on curated, edited, fact-checked long-form content. The quality ceiling of RAG is the quality of the corpus; a Gemini grounded on O’Reilly’s technical catalog or JHUP’s academic list is working from better material than one grounded on the open web.

For publishers, it’s a live experiment in whether licensing beats litigation. The industry spent 2023–2025 suing AI companies (or being sued by them) over training data. Expert Intelligence is the counter-model: keep training and querying separate, pay through the existing purchase funnel, and treat AI interaction as a feature of ownership rather than a replacement for it. Whether the “discovery loop” actually generates incremental sales — or cannibalizes them by letting readers skim a book through a chatbot — is now measurable.

For Google’s competitors, it raises the licensing bar. OpenAI and Anthropic have publisher deals of their own, but none has wired licensed books into a consumer note-taking product at six-figure title scale with ownership enforcement. Amazon is the obvious pressure point: Kindle owns the e-book market, and Google just made Play Books — the distant runner-up — the only e-book store whose purchases work as AI sources. Amazon’s response, whether a Kindle AI reader or deeper Alexa/Knox integration, is now a when-question.

For authors, the Featured Notebooks signal a new craft: authors who curate their own AI companions may find the notebook becomes part of the book’s value — an updating, interactive edition that keeps readers inside their intellectual world.

The open questions are just as sharp. Cited answers only cover queries; nothing stops a determined reader from extracting a chapter’s structure through patient questioning. The ownership check binds to Google accounts, not people, so household sharing gets awkward. And the feature is English-first, U.S.-weighted, and Play Books-only at launch — a fraction of the world’s readers own their books in that one store.

But as a template — licensed corpus, enforced ownership, citations back to the source, artifacts that study the material rather than replace it — Expert Intelligence is the most publisher-friendly AI feature any major lab has shipped. The book industry spent three years asking “what does AI do for us?” Google’s answer, shipping now in Gemini Notebook, is: it makes the books you already sold more useful to the people who already bought them.