Patna: Anthropic has launched a major update to its AI chatbot, Claude, allowing it to perform tasks on a user’s computer independently. The new functionality enables Claude to open files, navigate web browsers, use applications, and run development tools without direct human intervention.
The capability, currently offered as a “Research Preview,” is available to Claude Pro and Max subscribers on macOS devices. It integrates with Anthropic’s existing Claude Code and Cowork AI tools, building on earlier autonomous features introduced in the Claude 3.5 Sonnet model earlier this year. The company has now refined these features into specialised agents aimed at programmers, while still supporting general users.
To use the feature, users must keep the Claude desktop application running and pair it with the mobile app. Claude first attempts to complete tasks through direct connectors, such as Slack or Google Workspace. If no connector is available, it can operate the computer directly—controlling the browser, keyboard, mouse, and screen—though Anthropic emphasises that it will always seek explicit permission before performing any action.
The update also works in tandem with “Dispatch,” a cross-device tool that allows tasks to be assigned from a smartphone. Whether drafting a report, checking emails, or preparing a briefing, Claude can continue working on a user’s computer even when they are away.
Anthropic cautions that the feature is still in development. Complex tasks may require retries, and execution via the screen interface is slower than through direct system integration. Nevertheless, the capability offers early promise for automating routine office work and streamlining programming sessions.
With this update, Claude is evolving beyond a chatbot into a digital assistant capable of managing tasks autonomously while maintaining user privacy. Anthropic plans to expand support to Windows devices in the future, broadening access beyond the current macOS-exclusive release.
