Gemini Notebook

Google LLC

About this app

Turn complexity into clarity with Gemini Notebook, your brain’s new best friend. Join the millions of students, creators, researchers, professionals, CEOs, and more who are saving time, getting stuff done, and learning in new ways.

“One of the most compelling and completely flabbergasting demonstrations of AI’s potential yet.” – The Wall Street Journal

Now, with the Gemini Notebook app, you can create and access notebooks, ask questions whenever you think of them, and listen to your podcast-style Audio Overviews on the go with background playback and offline support.

📚 UPLOAD SOURCES
Upload all your long and complex PDFs, websites, YouTube videos, or text into a notebook.

💬 INSIGHTS YOU CAN TRUST
Gemini Notebook becomes an expert on your sources, summarizing them and making interesting connections. Then, you can ask it questions about, well, anything — and you can trust the answers, since your sources are cited in-line.

🎧 LEARN ON YOUR TERMS
Long blocks of text not your preferred way to learn? Transform what you uploaded into something more your speed, like a podcast-style audio discussion with two engaging AI hosts. You can even join the show to ask questions or steer the conversation in a different direction.

Ratings and reviews

I know it's not meant to represent the full experience, but you cannot do anything in the studio other than generate AI voice overviews. Why can I not see or edit other notes on my studio while using the app? It's exactly where I want to make quick notes for future activities when I'm in the full browser version. That aside, I love NotebookLM as a tool. Flipping brilliant.

Excellent tool for studying a subject, with known sources (maximum 50 in free version), thus reducing risks of hallucinations. Unfortunately however, it has recently lost the ability to sort notebooks between either "Most Recent" or "Title": I now have to use the "Search" feature to select the notebook needed for most new "Sources".

Being "encouraged" to use this for work. So far it struggles l prepared the supporting information, instructions and prompts to establish context but ultimately it took me longer than it would have to just do the work myself as it required so many corrections. About 1/6 of the output was usable. I repeated the same process with Mistral AI and Perplexity and these both required significantly less effort. Perhaps the use case was too specific or they "borrowed from different websites to train.

What’s new

Bug fixes and performance improvements.