Google NotebookLM has survived the AI product cycle longer than most Google tools. It launched in 2023 as a grounded research assistant: you feed it your own documents, it answers questions based only on those documents, and it tells you exactly where each answer comes from. That makes it different from regular AI chatbots, which can guess or invent facts. NotebookLM does not invent. It sticks to what you gave it.
Over the past two years it has grown steadily. The Audio Overviews feature, which turns your documents into a short podcast-style conversation between two AI voices, became genuinely popular. Slide decks, infographics, mind maps, and data tables followed. By late 2025 NotebookLM was running on Gemini 3 and had been formally built into Google Workspace plans for business teams.
Today's update has two parts worth understanding separately.
The first is the model change. NotebookLM now runs on Gemini 3.5 Flash. Google tested it against the previous version across five task categories and the new model won more than 65% of the comparisons. The practical meaning: the answers are more accurate, it handles documents in more languages, and it processes complex multi-document research better than before.
The second part is the Antigravity integration, and this is the more consequential shift. Antigravity is Google's coding platform, launched in late 2025. It allows AI agents to write and actually run code, not just suggest it. With Antigravity built into NotebookLM, each notebook now has access to what Google calls a "secure cloud computer." That means NotebookLM can now write code behind the scenes to do calculations, combine data from different sources, build charts, and export polished reports.
This is a clear step up from summarizing. Before today, if you uploaded a spreadsheet of sales figures, NotebookLM could describe the data. Now it can process the data, do the math, and hand you a finished analysis. You do not need to know anything about code for this to work. You describe what you want, and the system handles the rest.
The more than 100 pre-built skills that come with the Antigravity connection give NotebookLM a toolkit for specific tasks: deeper web research, structured data analysis, and richer formatted outputs. Google gives the example of a data analyst combining inconsistently formatted reports from multiple countries, asking NotebookLM to find context from the web, run the actual calculations, and produce a PDF report. The scenario is realistic and would previously have required several separate tools and probably a developer.
For business teams, the practical upgrade path here is clear. Research-heavy roles, from competitive analysis to procurement review to compliance documentation, stand to compress their workflows noticeably. The tool also now guides you through building a source library from scratch if you are starting with just a rough idea, and it can pull in relevant sources directly from Google Search.
There is a catch on timing. The full Antigravity-powered experience is rolling out now for Google AI Ultra subscribers and Workspace business customers with AI Ultra access. Everyone else on free or standard plans will get it later, with no specific date given.
NotebookLM has the rare quality of being genuinely useful for non-technical professionals without requiring any technical background. This update extends that usefulness from reading and summarizing documents to actively working with the data inside them.