Computer use: Computer use is an AI capability in which a model operates a computer the way a person does: looking at the screen, moving the cursor, clicking, typing, and switching between apps and web pages to finish a task in software that has no direct integration.
Most AI integrations work through APIs, structured connections that let software talk to software. Plenty of the tools people actually work in have no usable API: a vendor portal, a legacy desktop app, a government filing site, an internal system built in 2009. Computer use bridges that gap by letting the model take screenshots, interpret what it sees, and act with simulated mouse and keyboard input, looping until the job is done or it gets stuck.
Anthropic introduced a computer use capability in late 2024, and OpenAI, Google, and others have since shipped their own browser and desktop agents. As of mid-2026 the capability is real but uneven: agents handle routine form-filling and data lookup reasonably well, and they still misclick, get lost in pop-ups, and move far slower than a person on tasks a human finds trivial. Check the vendor's current documentation for what each product supports.
Treat it as powerful and risky. An agent that reads web pages is exposed to prompt injection: text on a page can try to redirect it. An agent with your logged-in browser can send, buy, or delete. Run it in a separate browser profile or sandbox, give it accounts with the minimum permissions, and require a human confirmation before payments, submissions, and anything that sends a message. Watch the first several runs end to end.
Example at work
An insurance agency needs claim status from a carrier portal that offers no API and no export. A computer-use agent logs in with a read-only account, searches each of forty claim numbers, and copies the status and last-updated date into a spreadsheet. A staff member reviews the sheet and handles the three claims the agent flagged as "could not find".
Why it matters
Computer use is how AI reaches the many systems that will never get an API, which is where a lot of tedious office work lives. It also puts an AI at your keyboard, so the permission and review rules matter more here than anywhere else.
Related terms
- AI agentAn AI agent is a system that uses an AI model to pursue a goal on its own by planning steps, using tools such as web browsers, files, and software APIs, and adjusting based on results, rather than answering a single prompt.
- Agentic AIAgentic AI describes AI systems designed to act with autonomy: they take a goal, break it into steps, use tools, and carry the work through with limited human direction. It is the broader approach; an AI agent is a specific system built that way.
- Tool useTool use (also called function calling) is the ability of an AI model to request that an outside tool be run, such as a web search, a calculator, a database query, or a calendar lookup, and then use the result in its answer.
- Prompt injectionPrompt injection is an attack in which instructions hidden inside content an AI system reads, such as a web page, email, or document, hijack the system into doing something its user never asked for.
- Robotic process automation (RPA)Robotic process automation (RPA) is software that mimics a person's clicks and keystrokes to run repetitive, rule-based tasks, such as copying invoice data from email into an accounting system. It follows a fixed script and makes no judgment calls.