On July 16, 2026, Google renamed NotebookLM to "Gemini Notebook." It remains an independent research tool, but the same notebooks now sync with the Gemini app, and going forward they will also be accessible from Google Search's AI Mode. At the same time, Google is extending to Google AI Pro users over the coming weeks an update to Gemini 3.5 and Antigravity that adds code execution to the tool's traditional function of reading documents and answering questions. Behind the name change lies a mechanism for carrying documents, instructions, and conversations across Google's AI products.
NotebookLM has distinguished itself from typical AI chat by answering based on documents the user selects and showing references back to the source text. That boundary still holds. However, on the Gemini side, users can now also draw on web search and various tools, while on the Gemini Notebook side, users can run code to produce deliverables. As convenience increases, it becomes more important to identify which screen is being used, what grounds an answer, and where the data ends up.
Connecting NotebookLM's 30 Million Users to Gemini
Google unveiled Project Tailwind at Google I/O in May 2023. That July, it was renamed NotebookLM and began as an experiment where users could select Google Docs as a basis for summaries and Q&A. The name "LM" reflected the design concept itself: a notebook built around a language model.
Three years on, its reach has expanded. According to Google, it now has more than 30 million users across over 600,000 organizations. Even after this renaming, Gemini Notebook will remain an independent product and continue to serve as the central tool for research. This is not an announcement that folds NotebookLM into the Gemini app and shuts it down.
What is changing is the scope in which notebooks can be used. In April 2026, Google introduced "notebooks" within the Gemini app, allowing users to store documents for complex projects, past conversations, and custom instructions all in one place. The company calls this a "personal knowledge base" shared across Google products. Add a document in the Gemini app and it appears in Gemini Notebook too, with name and instruction changes reflected bidirectionally.
July's renaming aligned the product name with this unification. Google further revealed plans to make existing notebooks available through Search's AI Mode as well. However, no release date has been announced, and it remains unclear whether documents can be added or edited from AI Mode, or how sources will be displayed there.
From a Reading Tool to an Execution Tool, with Over 100 Skills
The functional shift began in June. Google updated Gemini Notebook's foundation to Gemini 3.5 and Antigravity, assigning each notebook a "secure cloud computer." The AI writes code using the documents in the notebook and executes it on the spot. Google says it has also prepared more than 100 curated software skills, enabling everything from organizing inconsistently formatted data to performing calculations and turning results into charts or reports, all within a single conversation.
The range of files that can be created has also expanded beyond traditional summaries and study materials to business software formats.
| Use Case | Main Formats |
|---|---|
| Charts/Images | PNG, SVG, JPG, GIF |
| Documents | PDF, DOCX, Markdown, TXT |
| Structured Data | CSV, JSON |
| Business Files | XLSX, PPTX |
Generated files can be downloaded from the Studio panel, and users can request edits after creation. For example, one could envision loading statistics with mixed notation across multiple countries, using code to standardize units, and producing a chart and a PDF report. Source-attributed reading comprehension, which NotebookLM has excelled at, is now connected to calculation and file generation.
In Google's internal comparisons, the new system achieved an average win rate of over 65% against the old system across five major categories. Large-scale document analysis scored 69.9%, while advanced web research and document discovery scored 78.2%. The evaluation covered source-attributed Q&A and multilingual support, as well as long-document comprehension, deliverable creation, and research using multiple documents. However, this is a relative evaluation against the previous system, not an absolute accuracy rate for answers. Nor is it a third-party verification result, so how much it actually reduces corrections in practical use needs to be confirmed separately.
If code is going to run in the cloud, an explanation of the execution environment is essential. Google describes it as "secure," but has not disclosed details of the isolation method, the scope of network access, execution time limits, or audit log specifications. The level of management required differs between formatting spreadsheets and analyzing internal company data. As the rollout to Pro proceeds, this information will factor into usage decisions.
Same Notebook, Two Answer Modes
Even with a unified brand, Gemini Notebook and the Gemini app do not behave identically. According to Google's help documentation, chats in Gemini Notebook answer based solely on the documents within the selected notebook. Users can reference the cited passages in the source text, and can toggle whether each document is included in answers. This mode suits situations where you want to keep the scope of investigation closed and verifiable.
On the other hand, opening the same notebook in the Gemini app lets you use web search and Gemini's tools in addition to the saved documents. You can also move past Gemini conversations into the notebook and sync that conversation as new context. This is powerful when searching for the latest information beyond the documents or handing work off to other Google features, but it means an answer is not necessarily generated solely from the documents in the notebook.
Responsibility for deliverables is also divided. Studio's generation features—Audio Overview, Video Overview, infographics, and slide decks—remain on the Gemini Notebook side and cannot be created from the Gemini app. In other words, the design separates a screen for document-only verification from a screen for work that includes external search and tools, both operating on the same knowledge base. Even with the name unified, the scope of sources does not automatically become the same.
With the June update, Gemini Notebook itself gained the ability to search for candidate documents using Google Search. Here, the user reviews documents found by the AI and chooses which ones to adopt into the notebook. Attribution remains after documents are incorporated. The fact that it allows web research while returning the decision of what to add to the knowledge base to the human is what distinguishes it from a typical search-enabled chat.
Expanded Pro Rollout and Remaining Account Barriers
The update including code execution began in June with Google AI Ultra and Workspace enterprise users with AI Ultra Access or AI Expanded Access. In the July 16 announcement, Google stated it would roll out to all Google AI Pro users on the web version over the coming weeks. It will not reach all Pro accounts simultaneously on the announcement date.
Usage limits also vary by tier. According to the official table for individual users, Standard allows 100 notebooks, 50 documents per notebook, and 50 chats per day. Pro increases this to 500 notebooks, 300 documents, and 500 chats per day. Ultra allows up to 600 documents and up to 5,000 chats per day. Google explicitly states that these limits are subject to change, and the same table does not specify execution time or frequency limits for the cloud computer.
Account types are also not uniform. Gemini Notebook itself is available on most work and school accounts, but notebooks within the Gemini app are currently intended for personal Google accounts and do not support work or school accounts. Shared Gemini Notebooks also do not appear in the Gemini app's list. While individual research can now take advantage of cross-product syncing, projects involving organizational collaboration remain on the Gemini Notebook side.
Data handling also has boundaries. Google explains that files, generated content, and chat history added to Gemini Notebook are not used to directly train foundation models, except when users submit feedback. However, data shared to the Gemini app is also subject to the Gemini Apps privacy notice. Turning off "Keep Activity" in the Gemini app or deleting Gemini Apps Activity does not delete data stored within Gemini Notebook. It is not designed so that deletion and retention can be managed through a single setting.
How Far Will Search AI Mode Carry Over Notebooks?
The renaming to Gemini Notebook makes Google's intentions easier to see. Notebooks that store documents are being connected to Gemini's long-term projects, with computation happening in the cloud and results finished as documents or spreadsheets. In the future, the same notebooks will be usable from Search's AI Mode as well. The plan is to keep the independent tool used by 30 million people intact, while making its knowledge callable from Google's major AI surfaces.
At the same time, the boundaries between products become harder for users to see. Gemini Notebook is document-limited, while the Gemini app combines the web and tools, and the two differ in their history retention settings as well. Once Search's AI Mode is added, there will be a need for indicators showing which answers are based on saved documents and which information was supplemented by search.
What's worth confirming during the coming weeks of Pro rollout is how stably the cloud computer can handle large spreadsheets and lengthy documents, and whether users can trace the executed code and calculation process. And for Search integration, whether the right to choose sources and the right to delete are preserved will be key criteria for judgment. Only once these two points are demonstrated can Gemini Notebook be evaluated as a research platform that goes beyond mere brand unification.