Computer Use
In one line
Computer Use is the agent capability of perceiving the screen and driving the mouse and keyboard, letting an AI operate a computer the way a human would.
Going deeper
Computer Use jumped into the spotlight when Anthropic shipped it as a Claude capability in late 2024. The point is that the agent can drive any application — even ones with no API — by reading screenshots and operating the mouse and keyboard.
For marketers it opens scenarios where ad platforms, CRMs and dashboards designed for humans can now be operated by agents. Think of it as RPA, but with an LLM doing the reasoning instead of brittle scripted recorders.
The cost is reliability and security overhead. Screen parsing is imperfect, a wrong click can do real damage, and permissions need careful design. Production deployments almost always run inside an isolated VM or sandbox, with human approval gates at high-impact steps.
Sources
Related terms
Browser-use Agent
A browser-use agent is an agent that drives a real web browser — navigating sites, filling forms, even completing checkouts — and is becoming a default interface in AI search and shopping.
AI AgentAI Agent
An AI agent is an LLM-driven system that takes a goal, plans the steps, calls the tools it needs and runs the task end-to-end with limited human input.
AI AgentSandboxing
Sandboxing means running an agent in an isolated environment so its actions cannot reach the outside system — a baseline practice for any autonomous agent.
AI AgentPermission Model
A permission model defines which tools, data and actions an agent is allowed to touch — the core safety layer for any autonomous agent.
LLMClaude
Claude is Anthropic's LLM family, known for safety alignment, long-context handling and strong tool use — widely adopted in enterprise and developer settings.
How does your brand show up in AI answers?
Villion measures how your brand appears across ChatGPT, Perplexity and AI Overviews, then automates the work that lifts citation rate and share of voice.
Get a free audit