Home / Technology / Chrome's AI Agent Takes Over Web Tasks
Chrome's AI Agent Takes Over Web Tasks
18 Feb
Summary
- Gemini AI performs multi-step tasks across the web for users.
- Requires latest desktop Chrome, Google account, and AI subscription.
- AI agent requires user permission for sensitive actions and sign-ins.
Google Chrome has introduced 'Auto Browse,' an ambitious AI agent feature leveraging Gemini to perform multi-step tasks across the web. This tool allows the AI to navigate, search, compare, and even complete purchases on behalf of eligible US desktop subscribers. Users initiate tasks by describing their needs in a side panel, with Gemini then handling the browsing and providing prompts for confirmation.
Auto Browse integrates with other Google applications like Docs and Gmail, requiring user permission for access. It pauses to confirm sensitive actions such as sending emails or making purchases, and can optionally use Google Password Manager for site sign-ins. The feature requires the latest desktop Chrome version, a personal Google account with Gemini enabled, and either a Google AI Pro or Ultra subscription.
While powerful, Auto Browse requires explicit user permission for many actions and cannot handle follow-up tasks mid-stream, necessitating new prompts or manual intervention. Google AI Pro subscribers are limited to 20 multi-step tasks daily, while Ultra subscribers receive 200. Compared to other agentic browsers, Auto Browse is seen as a supplementary feature within Chrome's existing interface.
