Google has launched a new Expert Intelligence feature for its AI-powered note-taking service Gemini Notebook. Users can now load purchased books from Google Play Books directly into the app to ask questions about their content, generate infographics, podcasts, planners, or educational quizzes.
Google has launched a new Expert Intelligence feature for its AI-powered note-taking service Gemini Notebook. Users can now load purchased books from Google Play Books directly into the app to ask questions about their content, generate infographics, podcasts, planners, or educational quizzes.
The developers announced this on the Google blog. At the initial stage, the Expert Intelligence feature supports more than 100,000 e-books from leading publishers, including Penguin Random House, Johns Hopkins University Press, Macmillan Publishers and O’Reilly Media.
The new feature allows you to combine expert knowledge with your own materials. For example, managers can use management literature to prepare for difficult conversations with the team, and developers or analysts can use technical literature to analyze projects.
The developers have also built in copyright protection mechanisms. If a user shares a chat with a downloaded book, other participants will see it in the list of sources, but will not be able to view the text or generate replies based on it until they purchase their own copy from Google Play Books.
In the future, Google plans to expand Expert Intelligence to work with academic articles, periodicals, and business research reports, as well as integrate this tool into Search and the Gemini app.
Gemini Notebook is Google’s experimental AI tool for organizing and analyzing information. The service allows you to upload a variety of documents and sources, create summaries, audio reviews, diagrams, and answer questions about the content of the materials provided.
Google previously introduced Gemini 3.5 Transcribe, a new speech recognition model that not only translates audio into text, but can also structure chaotic spoken language. The model supports over 85 languages, recognizes specific terms, and can remove parasitic words.
