Google is adding Deep Research to its AI-powered writing and research assistant NotebookLM, adding agentic capabilities that transform it to an autonomous research agent that can find and evaluate information on its own. Previously NotebookLM was limited to retrieval-augmented generation (RAG). Now, led by your questions, Deep Research browses hundreds of websites, refining its search as it learns, generating within minutes an organized, source-grounded report that can be added to your notebook. Google is also adding support for more file types, including photos — allowing you to upload snapshots of notes — Google Sheets and Microsoft Word documents.

Also added are PDFs from Google Drive and the ability to add drive files as URLs with a copy-paste of the address for any website or YouTube video. This includes multiple links, separated by commas.

In a blog post that explains how to use the new features, Google explains “the report is just the beginning.” You can drop the compiled reporting straight into a notebook while continuing to add to it, using NotebookLM’s Audio and Video overviews to review big ideas, assembling “a rich knowledge base on any topic without leaving your workflow.”

Digital Trends says “NotebookLM is edging closer to being a proper research home” by integrating “the files and services you already use instead of asking you to rebuild everything from scratch” by “copying chunks into a separate tool.”

After opening NotebookLM and selecting the web in the source panel, users can choose between Fast Research for a jumpstart focused on core sourcing or Deep Research for a full briefing that compiles in the background as you continue to work.

“Since its launch in late 2023, Google has been building out NotebookLM,” reports TechCrunch, noting that “earlier this year, Google introduced Video Overviews to NotebookLM to allow users to turn dense multimedia, such as raw notes, PDFs, and images, into digestible visual presentations.”

The enhancement builds on the Audio Overviews capability, which can output AI podcasts “based on documents shared with NotebookLM, such as course readings or legal briefs,” TechCrunch adds.

In May, Google released iOS and Android NotebookLM apps, taking the service beyond desktop use via the web.