TL;DR: Google is rolling out Deep Research agents for NotebookLM, enabling automated synthesis of complex online research. The update also adds support for Google Sheets, images, Word documents, and direct Drive file integration.
Google has announced significant enhancements to NotebookLM, its AI-powered research platform, introducing automated research capabilities and broadening the range of supported file formats. The updates aim to streamline knowledge gathering whilst maintaining the platform’s source-grounded approach.
Deep Research Automation
The new Deep Research feature functions as a dedicated AI researcher, creating research plans and browsing hundreds of websites to generate organised, source-grounded reports within minutes. Users can direct the agent to search specific locations and refine queries as research progresses.
NotebookLM distinguishes itself by treating generated reports as starting points rather than final outputs. Users can import both reports and their underlying sources directly into notebooks, combining automated research with additional manual sources. The platform’s existing capabilities—including Audio and Video overviews—can then be applied to the assembled knowledge base.
Two research modes are available: Fast Research for quick scans with immediate source review, and a more comprehensive option for complex inquiries. Deep Research operates in the background, allowing users to continue adding sources without interrupting workflow.
Expanded File Format Support
Based on user feedback, NotebookLM now accepts Google Sheets for structured data analysis, images of handwritten notes or brochures, Microsoft Word documents (.docx), PDFs directly from Google Drive, and Drive files via URL with simple copy-paste functionality supporting multiple comma-separated links.
These features will roll out to all users over the coming week, with image support arriving in subsequent weeks. The expanded file type support addresses a common pain point for researchers working across multiple document formats and platforms.
Source: Google Blog