Anthropic Transforms Claude into an AI Agent

Anthropic has officially upgraded its AI chatbot, Claude, into an AI agent capable of utilizing a computer to perform various tasks.

Users can now enable Claude to operate their computers, opening applications, navigating browsers, and even filling out spreadsheets, mimicking tasks typically done at a desk. This functionality is currently available in the Claude Cowork and Claude Code applications for macOS only.

— Claude (@claudeai) March 23, 2026

The AI agent allows users to send commands from their mobile devices, enabling seamless task management. For instance, one user successfully instructed Claude to export a presentation to PDF and attach it to a meeting invitation while they were running late.

“Just send a prompt from your phone and return to the completed work on your computer. Claude can check your email daily or generate weekly reports,” Anthropic explains on its website.

While this computer functionality marks a significant step forward, Anthropic cautions that it is still in a preliminary phase compared to Claude's established skills in programming and text generation.

“The chatbot can make mistakes. We are continuously working on safety measures, but threats are always evolving,” representatives from Anthropic stated.

To mitigate risks, Claude requires permission to access new applications and presents a plan before executing any task. Users can also modify the task or select alternative methods throughout the process.

This new service is accessible through the Claude Cowork and Claude Code platforms, exclusively for macOS users. Notably, Claude recently achieved the top position among free apps in the US App Store, surpassing OpenAI's ChatGPT.