Google rolls out Gemini in Chrome to US users with agentic browsing
Beyond summarising and comparing open tabs, Google said Gemini would soon complete tasks like booking a haircut or checking out a grocery order without further input.
- Models & capabilities
- Minor
Google made its Gemini assistant available as a built-in Chrome feature to all Mac and Windows desktop users in the US with English-language settings, accessible from an icon in the browser’s toolbar. The assistant could read and summarise the content of the open page, compare and summarise information across multiple open tabs at once, and — in an upcoming update — search a user’s browsing history in natural language rather than requiring them to recall a specific page.
Google said further “agentic” capabilities were coming in the following months: Gemini navigating to a website, adding items to a cart and letting the user complete the final checkout, with examples given of booking a haircut or ordering groceries. Chrome also gained an AI-powered address bar mode for follow-up search queries and a Gemini Nano-based scam detector for identifying fake alerts and phishing pages, alongside one-click password resets on supported sites.
The rollout put Chrome’s several hundred million US users into a live testbed for agentic browsing — AI systems that act on a user’s behalf inside a browser rather than only answering questions — a capability OpenAI, Anthropic and Perplexity were pursuing in parallel through separate browser and browser-extension products around the same period.