MOUNTAIN VIEW, Calif. — January 2026 — Google is bringing more agentic capabilities to the browser with Gemini 3 Auto Browse, a new feature that allows AI to navigate websites, click through pages, and complete tasks directly within Google Chrome.
Auto Browse is powered by Gemini 3 and is designed to handle multi-step web tasks that typically require manual effort. Instead of users switching tabs, scrolling pages, and filling forms themselves, Gemini can now take action in the browser while explaining what it’s doing along the way.
Google positioned Auto Browse as a way to turn Chrome into an active assistant rather than a passive tool. Users can ask Gemini to research information across multiple sites, compare options, summarize findings, or carry out repetitive web actions, all while staying in control of the process.
The feature works by letting Gemini understand page layouts, buttons, forms, and navigation elements visually and structurally. This allows the model to move through real websites much like a person would, adapting as pages change rather than relying on fixed scripts.
Google emphasized that Auto Browse is designed with transparency and user control in mind. Gemini shows each step it plans to take and asks for confirmation before performing sensitive actions. Users can interrupt, adjust instructions, or take over at any time.
Auto Browse reflects Google’s broader push toward agentic AI, where models don’t just generate text but actively perform tasks across tools and environments. By embedding these capabilities directly into Chrome, Google is aiming to reduce friction between intention and action on the web.
Gemini 3 Auto Browse is rolling out gradually as an experimental feature, with Google continuing to refine safety controls, reliability, and supported use cases as feedback comes in.
Source: Google
Join the conversation! Your thoughts help the community grow.