MOUNTAIN VIEW, Calif.: Google has given NotebookLM a major upgrade, powering it with Gemini 3.5 and new agentic research capabilities that transform it from a simple document summarizer into an active research partner.
The update rolled out globally on June 8 for AI Ultra and Google Workspace customers. With the new version, NotebookLM can now write code using Antigravity-powered coding skills, independently discover sources, and automatically build entire project repositories from scratch without any manual input.
The upgrade marks a significant shift in what NotebookLM is designed to do. Instead of just summarizing documents that users upload, it now actively finds information, organizes it, and helps users build research projects on its own.
Gemini 3.5 replaces the previous model as the default reasoning engine across NotebookLM. The upgrade improves accuracy, large-document handling, and multistep reasoning across complex projects. Users now see the AI’s thinking process transparently inside each notebook session.
Every notebook gains access to a secure cloud computer that executes code on demand. The sandbox runs Python and ships with more than 100 built-in software skills. Tasks include data analysis, chart generation, statistical work, and document transformation workflows.
The new agentic research framework lets users start with just a topic or question. NotebookLM autonomously crawls the live web for primary sources with explicit user permission. The system unearthes cross-language references and targeted reference works for the project.
NotebookLM now generates downloadable outputs including PDFs, spreadsheets, charts, images, and presentations. The expanded format support targets professional research workflows beyond casual note-taking. Output flexibility positions NotebookLM closer to a full research workstation than a chatbot.
The launch follows Apple’s WWDC 2026 announcement that Siri AI runs on Gemini models. Google’s research tool now competes more directly with OpenAI’s Deep Research and Anthropic’s similar agentic offerings. Each major lab races to make research a core agentic AI use case.
Google plans broader access and additional output formats over coming months. The update arrives the same day Google cuts its AI Plus consumer pricing significantly.
Previous
