NotebookLM Becomes Gemini Notebook — and Gets a Code Sandbox

Google is retiring the NotebookLM name. As of July 16, the AI research-and-note-taking app is Gemini Notebook — part of a broader push, announced by Josh Woodward, VP of Google Labs, Gemini app & AI Studio, to pull Google's scattered AI products under one brand. It stays a standalone app rather than folding entirely into Gemini, but it now syncs with the Gemini app directly, and Google says notebooks are coming to AI Mode in Search soon.

The rename is the least interesting part

What actually changes the product is a new "secure cloud computer" attached to every notebook. It lets Gemini Notebook write and execute code natively against a user's uploaded sources — meaning you can ask it to run real analysis on your documents, not just summarize or quote them, and get back new output formats rather than static prose. That's a meaningful shift from 'chat with your PDFs' to 'compute on your PDFs.'

The rollout is staged: available now for Google AI Ultra subscribers and Workspace business customers, with general Pro-tier availability coming over the next few weeks.

Why builders should care

NotebookLM's core trick was always source-grounding — answers tied to documents you explicitly gave it, with citations, instead of open-ended model recall. That's the same property that matters for anyone doing AI-assisted development: you want an agent reasoning over your actual codebase, spec, or dataset, not its training-data guess at what your codebase probably looks like. Wiring in a sandboxed code-execution layer on top of that grounding is a small but real signal about where the baseline for 'AI research tool' is heading.

• Grounded execution over open recall — running code against sources you control reduces the surface for confident-but-wrong answers, the same failure mode that matters when an agent is editing your repo.
• Cross-surface sync — the same notebook now follows you between the Gemini app and (soon) Search, which is the direction most serious dev tooling is also headed: one context, many entry points.
• Scale as validation — 30M+ users and 600,000+ organizations is a large enough base that Google is unlikely to walk this feature back; it's a reasonable bet as a stable part of the Gemini stack going forward.

A rebrand with a pattern

This is at least the third notable Google AI product to get folded into the Gemini name in the past year, and the logic is consistent: consolidate the brand, then use the unified surface to cross-sell features (Search, the Gemini app, Workspace) that used to live in silos. For a tool that started as Project Tailwind at Google I/O in May 2023 with a fairly narrow 'help people learn' mission, the code-execution addition is the more consequential move — it's Google admitting that grounded reasoning without the ability to compute on the data was leaving value on the table.