Chrome auto browse can help with work, says Google

Google is looking to expand Chrome’s role in enterprise productivity with a new auto-browse feature built on its Gemini 3 model that it says can navigate websites, gather information, and process it, reducing manual data entry and repetitive clicks in professional workflows.

The feature is available in preview to paying AI Pro and Ultra subscribers in the US through the Gemini interface in Chrome.

It arrives as hyperscalers and leading model providers such as OpenAI and Anthropic are pushing AI deeper into enterprise workflows, seeking to automate routine tasks and processes to deliver measurable productivity gains.

OpenAI made an early move in this direction in February 2024, showcasing software capable of operating devices autonomously, with Anthropic demonstrating its “computer use” capability in October that year. Since then, both companies have worked to fold these abilities into more refined products, either as standalone offerings or in Anthropic’s case, directly within Claude, its generative AI–based chatbot.

Google too has been experimenting with browser and agent-based automation for some time, including Jarvis, revealed in October 2024 and now called Project Mariner, which explores more autonomous web navigation and task execution as part of Google’s broader push to make AI an active participant in enterprise operations.

Jarvis was showcased just days after Anthropic showed off computer use and is now available to subscribers to the $250/month Google AI Ultra service as a prototype.

Available in preview

Chrome auto browse is available in the US as a preview for AI Ultra subscribers, and also for subscribers to the $20/month AI Pro service, a move that analysts say positions the browser as a lightweight productivity layer, aiming to streamline knowledge work and free employees from repetitive, time-consuming online tasks.

Avasant principal analyst Abhisekh Satapathy welcomed Google’s inclusion of user supervision capabilities, noting that Gemini asks for confirmation before completing certain actions.

Pareekh Jain, principal analyst at Pareekh Consulting, focused on its ease of use faced with complex workflows.

“It can handle complex multi-step web workflows like form filling and navigation, with enterprise use cases including expense processing (extracting receipts from portals), procurement quote aggregation across vendor sites, and CRM updates via SaaS interfaces,” he said.

Development teams could see productivity benefits, he said: “This could unlock substantial gains through zero-code automation, letting operations teams in HR or Finance create mini-automations independently such as instructing it to go to the vendor portal, download January invoices, and save to this Drive folder, without developer delays.”

Freed from such mundane work, Jain said, “Developers can then shift from crafting fragile web scraper scripts to authoring high-level agentic instructions, redirecting focus from clicks to desired outcomes, boosting efficiency across workflows.”

Everest Group practice director Priya Bhalla said Chrome auto browse could have a more profound impact on developers’ thinking about designing the user experience: “Over time, this could shift how developers think about UX — optimizing not just for human users, but also for AI agents acting on their behalf.”

Not without risks

However, the analysts cautioned that Chrome auto browse may not be well-suited for mission-critical workflows.

Enterprise systems involve authentication layers, role‑based controls, conditional logic, and custom interfaces, which are likely areas where Chrome auto browse could struggle, Jain said.

“It relies solely on browser interactions without deep API or internal system integration. Also, it could be brittle on dynamic web pages prone to DOM changes,” Jain added.

Typically, an agent uses Document Object Model (DOM) to navigate a webpage as it represents the structure of a webpage to locate buttons for clicks etc. In a dynamic webpage, the DOM might change frequently, creating challenges for the AI agent.  

Beyond reliability and integration challenges, analysts also flagged potential security risks associated with delegating browser-level autonomy to AI agents.

“These include handling authenticated browser sessions, interacting with untrusted external websites, and ensuring automated actions do not unintentionally submit incorrect or sensitive information,” Satapathy said. “In regulated environments, this can complicate audits and compliance reviews.”Chrome auto browse can help with work, says Google – ComputerworldRead More