Google used the opening day of Cloud Next 2026 in Las Vegas to push Chrome from a passive tab-and-bookmarks browser into what executives repeatedly called an "AI coworker." The centerpiece is Auto Browse, a Gemini 3-powered feature that can take over the keyboard and mouse to complete multi-step web workflows while the user does something else.
What Auto Browse actually does
Google's pitch is concrete. According to the company's Chrome blog, Auto Browse can plan a vacation by comparing hotel and flight costs across date ranges, schedule appointments, fill online forms, collect tax documents, renew licenses, pull quotes from service providers, manage subscriptions, hunt for apartments with custom filters, and shop within a budget while applying discount codes. It leans on Gemini 3's multimodal capabilities to read the live context of a user's open tabs, and it can tap Google Password Manager when the user authorizes it to sign in on their behalf.
On the enterprise side, TechCrunch reports that the same engine is being wired into Chrome Enterprise for knowledge-work use cases: CRM data entry, vendor price comparison, summarizing candidate portfolios, and competitor research.
Skills and a human in the loop
Google is layering a "Skills" concept on top of Auto Browse, letting users turn recurring prompts into one-click tools accessible from a forward-slash command or a plus-sign menu in the address bar. For the actions most likely to cause damage, the agent is explicitly designed to stop and hand control back to the human. Google's Chrome team writes that Auto Browse will "pause and explicitly ask for your confirmation" before purchases, email sends, or social posts — a hedge against the kind of rogue-agent incidents that have bruised rival browsers this year.
Enterprise controls and Shadow IT
For IT admins, the rollout comes with a tightened safety story. Auto Browse in Chrome Enterprise is gated behind an organizational policy, and Google states that "an organization's prompts won't be used to train its AI models." Alongside it, Chrome Enterprise Premium is getting Shadow IT risk detection that surfaces usage of both sanctioned and unsanctioned generative AI tools across an org, plus deeper integrations with Microsoft Information Protection for data classification and an expanded Okta partnership targeting session-hijacking attacks.
Availability
The consumer Auto Browse preview is limited to the US and to Google AI Pro and Ultra subscribers on Windows, MacOS, and Chromebook Plus, with Google describing a staged rollout rather than a single launch day. The enterprise version is flagged for US Workspace customers first, toggled on by administrators.
Why it matters
The agentic browser race has been crowded — Perplexity's Comet, OpenAI's Atlas, and Anthropic's Claude Computer Use have all made noise in the past six months — but none of them ships pre-installed on roughly three billion devices. By routing Gemini 3's agent directly into Chrome, Google is betting that distribution will matter more than novelty, and that its UCP-based commerce hooks with partners like Shopify and Target will turn browsing into a transactional surface agents actually own.
