Google officially retired the NotebookLM brand on Thursday, renaming the beloved research assistant to Gemini Notebook. The change pulls what was once an experimental side project directly into the core of the tech giant’s ecosystem.
But the name change is just window dressing for a much heavier technical leap. Over the coming weeks, Google is pushing its Gemini 3.5 model and the highly anticipated “Antigravity” upgrade down to standard AI Pro subscribers. Previously, this capability was strictly locked behind the AI Ultra tier.
Antigravity turns the platform from a passive reader into an active agent. Every Gemini Notebook is now provisioned with a secure, dedicated cloud computer environment. This lets the system write, execute, and run its own scripting code to analyze large document dumps natively. It generates entirely new output formats grounded strictly in the source material you upload.
The service is also breaking out of its standalone web page. Notebooks are already appearing inside the main Gemini chat interface, and they are slated to integrate directly into Google Search’s upcoming AI Mode very soon. That means active research sessions will follow you across different browser tabs without requiring a dedicated workspace.
The scale of adoption explains the aggressive push. Google confirmed the service now hosts over 30 million individual users and handles data for more than 600,000 organizations.
Why Localized Cloud Computing Beats Standard Prompting
We are watching the final evolution of a tool that started over three years ago as an experimental curiosity called Project Tailwind at I/O 2023. Back then, it was just a neat way to summarize PDFs. Now, by giving each notebook its own isolated computing environment, Google is changing how professionals handle raw information.
Instead of asking a language model to guess the math inside a 400-page financial report, the Antigravity agent writes a Python script, executes the code, and gives you the hard mathematical output. It is a fundamental shift from generating text to executing tasks. You are no longer chatting with a document. You are renting an automated data scientist.