NotebookLM just became an agent, not a reader, powered byGemini 3.5 and Antigravity

2026-06-22 · Source: AI on Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, long

Summary

Google's NotebookLM received a significant upgrade on June 8, 2026, transitioning from a document reader to an active research agent powered by Gemini 3.5 and Antigravity, Google's agent-first coding IDE. This update introduces four key engine-level changes: every notebook now includes a secure cloud computer capable of writing and running code with over 100 software skills; the chat can build source libraries from scratch using Google Search; outputs are downloadable in 12+ formats, including PPTX and XLSX; and NotebookLM now exposes its thinking steps in chat. This transformation allows for deeper research, complex analysis, and automated deliverable creation. Google's evaluations show an average win rate above 65% across core dimensions, with 69.9% on large document analysis and 78.2% on advanced web research. The update is rolling out on the web to Google AI Ultra and Workspace AI Ultra users first.

Key takeaway

For research scientists and data analysts building knowledge bases, NotebookLM's agentic upgrade changes how you approach complex projects. You can now initiate zero-upload literature reviews or process raw datasets directly within the tool, generating editable reports and spreadsheets. Be sure to rigorously vet auto-discovered sources and verify computed results, especially for high-stakes work, as the system's reasoning steps are a debugging aid, not a guarantee of correctness.

Key insights

NotebookLM's upgrade transforms it into an agent that discovers, analyzes, and delivers, powered by Gemini 3.5 and Antigravity.

Principles

• Agentic AI enhances research workflows.
• Visibility into reasoning improves trust.
• Code execution enables deeper data work.

Method

Start a blank notebook, ask it to discover sources via Google Search, curate results, analyze with the cloud computer, then export the deliverable in desired formats.

In practice

• Generate editable PPTX or XLSX files.
• Reconcile messy data formats programmatically.
• Build citation matrices from foreign sources.

Topics

• NotebookLM
• Gemini 3.5
• Antigravity
• AI Agents
• Cloud Computing
• Research Automation
• Data Analysis

Best for: AI Scientist, Research Scientist, Data Scientist, AI Product Manager

Related on AIssential

• NotebookLM Agentic AI Update Is HUGE! Agentic Coder Now? — NotebookLM's agentic update transforms it into a comprehensive research workspace with integrated cloud computing, source discovery, and artifact generation.
• Antigravity 2.0 UPDATE: NEW Agentic AI Coding Agent + Gemini Desktop App! — Google's Antigravity 2.0 modularizes AI development with autonomous agents and a dedicated SDK for custom solutions.
• Google's NotebookLM now runs its own cloud computer with code execution and agent-based research — NotebookLM now integrates cloud computing and agent-based AI to streamline complex research and content generation.
• How to Use NotebookLM with Gemini: A Practical Guide with 30 Use Cases — Integrating NotebookLM with Gemini combines source-grounded accuracy with broad AI capabilities for diverse applications.
• Do better research with NotebookLM — NotebookLM integrates advanced AI, code execution, and diverse output generation to streamline complex research from ideation to final reports.
• 🤖 AI Agents Weekly: Gemini 3.5 Flash, Antigravity 2.0, Codex Thursday, Cohere Command A+, Qwen3.7-Max, and More — Google's new Gemini 3.5 Flash and Managed Agents significantly advance AI agent capabilities and deployment ease.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI on Medium.