Anthropic has added computer use capabilities to the Claude Code CLI, letting the AI assistant open applications, click through user interfaces, debug native tools, and capture screenshots — all from the macOS terminal.
From Chat to Control
The feature, currently in research preview for Pro and Max subscribers on macOS, represents a significant expansion of what AI assistants can do. Rather than just generating text or code, Claude can now interact with your computer the way a human would — navigating graphical interfaces, clicking buttons, and completing multi-step workflows across different applications.
Computer use is accessed through the Claude Code CLI rather than the web interface, giving developers direct terminal access to Claude's new capabilities.
What Claude Can Do
With computer use enabled, Claude can:
- Open applications and navigate to specific features or settings
- Click through interfaces to complete tasks that require visual interaction
- Debug native tools by seeing what's actually on screen
- Take screenshots to verify results and show its work
- Chain together workflows that span multiple applications
The feature essentially transforms Claude from a text-based assistant into something closer to a robotic process automation tool, but with natural language understanding and reasoning capabilities.
The Agentic AI Trend
Anthropic's computer use launch comes as "agentic AI" has become the hottest category in the industry. The idea: AI systems that don't just answer questions but actually take actions on your behalf.
This week also saw Anthropic roll out MCP-connected work tools on mobile, letting users explore Figma designs, create Canva slides, and check dashboards from their phones. The combination of computer use on desktop and MCP tools on mobile positions Claude as an AI that can operate across your entire digital workflow.
Enterprise Implications
For developers and enterprise users, computer use opens new possibilities for automation. Tasks that previously required custom integrations or RPA tools can potentially be handled by Claude through natural interface interaction.
However, the research preview status signals that Anthropic is proceeding carefully. Letting AI control a computer raises obvious security and safety questions — and getting the trust and permission model right will be critical for broader adoption.
What's Next
Computer use is part of Anthropic's broader push to make Claude more capable of real-world tasks. As AI assistants evolve from conversation partners to autonomous agents, features like this will define who leads the next phase of the AI race.
The feature is available now for Pro and Max subscribers on macOS through the Claude Code CLI.
