Google upgrades NotebookLM into a more agentic research tool

Google has upgraded NotebookLM with a set of AI features that move the product beyond document Q&A and closer to a research workspace. In a post published on June 8, 2026, Google said NotebookLM is getting new agentic capabilities in chat, more advanced reasoning and a secure cloud computer that can write and run code for analysis tasks.

The update matters because NotebookLM has become one of Google's clearest examples of practical AI: a tool built around sources, citations and personal or organizational knowledge rather than an open-ended chatbot. The new version runs on Gemini 3.5 and Antigravity, according to Google, and gives each notebook access to more than 100 curated software skills. That allows the system to analyze data, work across large documents, discover relevant web sources and turn research into downloadable artifacts.

Google says NotebookLM can now generate more output formats directly from the studio panel. Users can ask for charts, reports, spreadsheets, documents, markdown files, structured data, images and slide decks, then edit the results after generation. That is a meaningful shift for researchers, students, analysts and teams who previously used AI mainly to summarize or explain materials. The product is now being positioned as something that can help produce finished work, not just notes.

The company also emphasized source discovery. Instead of starting only with uploaded files, users can begin with a loose question or idea. NotebookLM can then use Google Search to find relevant web sources and help build the notebook's source repository, while keeping the user in control of what gets added.

For AI users and businesses, the update shows where productivity tools are heading. Research assistants are becoming multi-step systems that can search, reason, calculate, create files and preserve attribution in one workflow. The rollout is global on the web for Google AI Ultra users and Workspace business customers with AI Ultra Access or AI Expanded Access, with broader availability planned later. That makes this less a small feature update and more a preview of how everyday knowledge work may be packaged inside AI products.