A book has traditionally been delivered as a finished object: print copy, ebook, audiobook, or file. The reader opens it, reads it, highlights it, and returns to it when needed.
AI is beginning to add another reading surface.
On August 27, 2026, Google introduced Expert Intelligence, a cross-Google initiative that lets users incorporate eligible trusted content into Gemini Notebook. At launch, the program includes more than 100,000 books from major publishers. For supported Google Play Books purchases, a reader can add the book to a notebook and ask questions grounded in that title, combine it with other sources, and generate learning artifacts such as quizzes, infographics, and audio overviews.
The important publishing signal is not that every book is suddenly inside an AI system. It is that a publication can increasingly function as a licensed knowledge source that readers interact with after the purchase.
The book is becoming a reusable knowledge object
Publishing has always created reusable knowledge, but the interface was mostly manual. A reader searched an index, reread a chapter, copied a note, or remembered where an idea appeared.
AI-native reading changes the interaction layer. The reader can ask a question in the context of the book, connect the book to personal information or other sources, and use the publication as a grounded reference inside a broader problem-solving session.
That changes the value of structure.
A well-organized publication with clear concepts, strong section hierarchy, specific terminology, examples, definitions, and evidence is not only easier to read. It is also easier to retrieve from, cite, compare, and apply.
Rights and access are part of the product
Google’s launch is notable because it is not presented as unrestricted ingestion of books. Google says users must own an eligible book through Google Play Books to interact with it through Gemini Notebook. If a notebook is shared with another person, that collaborator is prompted to purchase their own copy before they can use the book in the notebook; otherwise the book is unavailable to them.
That is strategically important for publishers.
The question is not only “Will AI summarize books?” The more useful question is “What does licensed, permission-aware interaction with a publication look like?”
If AI reading surfaces continue to expand, publishers will need to understand distribution rights, platform eligibility, metadata, licensing terms, attribution behavior, and how value is preserved when the content becomes interactive.
What this means for authors
For authors, the opportunity is larger than producing more pages.
An author who develops a coherent body of knowledge can become a source readers return to inside many different contexts. That favors books with durable concepts, clear models, useful language, well-supported claims, and practical application.
It also raises the value of author authority. If a reader can combine a book with their own notes, work documents, or other trusted sources, the author’s ideas have to remain distinguishable and reliable inside that larger context.
This is one reason weak, derivative publishing becomes less defensible as AI gets better at synthesis. A book that merely rearranges commodity information is easier to replace. A book with original frameworks, experience, research, examples, or a distinctive method has more durable source value.
What this means for publishers
Publishers should begin thinking about a title in three layers:
-
The reading product.
The book still needs to work as a book: coherent, edited, useful, well-designed, and satisfying to read. -
The source product.
The content needs enough structure, accuracy, metadata, and internal clarity to function as a dependable knowledge source when readers search or interact with it in new environments. -
The relationship product.
The title should connect the reader to a broader body of work: related publications, updates, supporting resources, services, community, or future editions when appropriate.
That does not mean stuffing books with machine-oriented markup or writing for an algorithm instead of a person. The best source structure is usually the same structure that helps a human reader: clear sections, explicit concepts, accurate references, useful summaries, consistent terminology, and a strong editorial hierarchy.
Do not assume automatic inclusion
Expert Intelligence is a specific Google program with specific participating content and eligibility. Its current launch should not be read as a promise that every independent title, PDF, or publisher can opt in immediately.
Google says more than 100,000 books from publishers including Bloomsbury, De Gruyter Brill, Johns Hopkins University Press, Macmillan, O’Reilly Media, and Penguin Random House are included at launch, along with Featured Notebooks developed with more than 15 bestselling authors. Google also says it plans to expand Expert Intelligence-enabled content to additional products and source types over time.
The strategic lesson is therefore directional, not speculative: major platforms are building rights-aware AI experiences around trusted, purchased source material.
How to prepare a publishing catalog for this direction
- Strengthen provenance. Make authorship, publisher identity, edition information, and source history clear.
- Improve structure. Use meaningful sections, descriptive headings, consistent terminology, and navigable references.
- Protect accuracy. Verify factual claims and distinguish evidence from opinion.
- Preserve rights records. Know what you own, what you licensed, what third-party material is included, and where distribution rights apply.
- Build durable metadata. Titles, descriptions, identifiers, categories, author information, and edition data should be complete and consistent.
- Create real intellectual value. Original frameworks, documented experience, and specific problem-solving methods are more defensible than generic synthesis.
- Plan for updates. In fast-changing subjects, edition control and update paths become part of trust.
Publishing is moving from files to systems
This shift fits a broader pattern. Readers increasingly encounter knowledge through search, AI answers, notebooks, audio summaries, interactive tools, and agentic workflows. The publication is still important, but it may no longer be the only interface through which its ideas are used.
That is why publishing strategy should focus on more than output volume. As discussed in Publishing More Is Not a Content Strategy, the durable advantage comes from building knowledge assets that compound rather than simply increasing the file count.
The practical takeaway
Expert Intelligence is an early example of a larger publishing direction: trusted content can become an interactive, licensed source inside AI experiences.
Authors and publishers should respond by improving the things that already make a publication valuable—originality, structure, provenance, rights clarity, accuracy, metadata, and a coherent body of knowledge.
For the broader publication-readiness framework, see Publish-Ready Book Build: What Publication-Ready Means and Mindset Media Group’s Publishing & Creation resources.
Source note: Google, Expert Intelligence: a new way for you to engage with trusted content, August 27, 2026. Google reports more than 100,000 participating books at launch and describes purchase-based access controls for eligible Google Play Books titles.
Related resources