What NotebookLM with Gemini 3.5 Now Is
NotebookLM with Gemini 3.5 is an AI note‑taking and research platform that combines source‑grounded chat, automated source discovery, and cloud‑based code execution to help students and professionals turn raw material into finished analysis, reports, and visual outputs. Google has rebuilt the chat experience on the NotebookLM Gemini 3.5 model family plus its Antigravity tool layer, giving the system stronger reasoning and clearer visibility into how answers are produced. Instead of starting from a pile of documents, users can now begin with a question or loose idea, ask NotebookLM to find relevant web sources, and curate which ones enter a notebook. Once sources are set, the AI can summarize, compare, and transform them while keeping explicit citations. This reframes NotebookLM from a passive summarizer into an active research partner that can plan workflows, perform calculations, and create exportable work products.
Agentic Research Tools: From Question to Curated Sources
The standout change is the move toward agentic research tools. NotebookLM can now respond to an initial question by using Google Search to identify candidate web sources, suggest related primary material, and uncover work by the same author or in other languages. Users still approve which sources are added, so the notebook remains grounded in a controlled library rather than an opaque web crawl. Powered by NotebookLM Gemini 3.5 and Antigravity, the chat interface exposes more of its reasoning steps directly in the conversation, giving researchers a clearer path to check how it reached a claim. According to Google, the upgraded system “achieved exceptional performance in advanced web research and source discovery reaching a 78.2% win rate against our prior baseline.” For students, research teams, and writers, this means the setup phase—finding and organizing readings—can be largely automated without giving up editorial control.
AI Code Execution and Secure Cloud Computers
Every notebook now includes its own secure cloud computer, introducing AI code execution into the research flow. Instead of exporting data to external tools, users can ask NotebookLM to write and run code that analyzes datasets, tests hypotheses, or generates charts based on the notebook’s sources. Google says the system ships with more than 100 curated software skills, so tasks like statistical analysis, simple simulations, or text processing can happen in one place. This capability shifts NotebookLM from static AI note‑taking into an environment closer to a programmable research assistant. In side‑by‑side evaluations against the previous version, Google reports an average win rate of over 65% across its top five evaluation dimensions, including a 69.9% win rate in large document analysis. Benchmarks are company‑provided rather than independent, but they hint at gains in both reasoning and code‑assisted workflows.
Research Automation, Outputs, and Workflow Integration
Once analysis is done, the upgraded NotebookLM focuses on research automation and export. The system can transform grounded sources into polished outputs such as PDF research summaries, written reports, slide outlines, spreadsheets, and charts. Android Authority notes that users can export documents in formats including PDF, DOCX, Markdown, plain text, and image formats like PNG and SVG, giving teams flexibility to move work into existing tools. Because outputs remain tied to the underlying sources, it is easier to audit claims or update a report when a notebook’s content changes. These capabilities matter for students preparing assignments, academics drafting literature reviews, and professionals turning briefings into client‑ready files. Instead of copying text between apps, the entire pipeline—from idea, to sources, to analysis, to final artifacts—can live inside NotebookLM and then flow outward through standard file formats.
Who Can Use It Now and What Comes Next
For now, NotebookLM’s biggest upgrade is limited to specific account tiers. The new NotebookLM Gemini 3.5 and Antigravity features are rolling out on the web to Google AI Ultra subscribers and Workspace customers with AI Expanded (also described as AI Ultra and Workspace access). Schools, universities, researchers, and workplace teams already using these AI offerings are expected to be among the first to try the agentic research tools and AI code execution. Google states that broader availability is planned, but has not given a detailed timeline or feature breakdown for free or lower‑tier plans. That means many users still see NotebookLM as an AI note‑taking tool, while early adopters can start treating it as a research automation environment. As access widens, the line between traditional notebooks, coding notebooks, and AI research agents is likely to blur further.
