API only: not available in the Claude app
What is Computer Use?
Computer Use gives Claude the ability to control a full desktop environment the same way a person would. It can see the screen, move the mouse, click buttons, type text, switch between applications, and complete multi step tasks autonomously. Think of it as remote access for an AI: you describe what you want done, and Claude operates the computer on your behalf.
This is one of the most advanced agentic capabilities in any AI system. While most AI tools are limited to generating text or images inside a chat window, Computer Use lets Claude interact with any software that has a visual interface. It opens, navigates, and operates applications just like a human user sitting in front of the screen.
Step 1: See the screen
Claude takes a screenshot and analyzes what is currently visible
Step 2: Decide next action
Based on the task, Claude determines what to click, type, or navigate to
Step 3: Execute the action
Moves the mouse, clicks buttons, types text, or switches between apps
Step 4: Verify and repeat
Takes a new screenshot, checks progress, and continues until the task is complete
What people use it for
- Automated testing: Navigate web applications, fill out forms, and verify that everything works as expected
- Data entry: Transfer information between systems that do not have APIs or integrations
- Repetitive workflows: Any multi step task that involves clicking through the same sequence of screens repeatedly
- Legacy software: Interact with older applications that can only be controlled through their visual interface
Computer Use vs. Cowork vs. Claude Code
- Computer Use (API): Full desktop control. Sees the screen, clicks, types. For developers building automation.
- Cowork (Desktop app): File focused. Reads and writes files in a folder you choose. Visual interface, no coding needed.
- Claude Code (Terminal): Code focused. Reads, writes, and runs code in your project. For developers who prefer the command line.