What is Computer Use?
Computer Use is the ability of AI models to directly control the computer interface: moving the cursor, clicking buttons, typing text, navigating websites and desktop applications. The model "sees" the screen (screenshot) and takes actions like a human.
How does it work?
A multimodal model analyzes a screenshot, recognizes interface elements (buttons, fields, menus), plans an action sequence, and issues commands. Between each step, it analyzes the new screen state and adjusts the plan.
Automation applications
Computer Use enables process automation in legacy systems without APIs: entering data in old applications, navigating supplier portals, filling administrative forms. It's the "last mile" of automation — where traditional API integration is impossible or uneconomical.