Checking written code on screen
Cursor, maker of an AI programming tool, added "Computer Use" on Oct. 23 (Oct. 24 Japan time), letting its AI agents operate desktop apps. The agents click on the screen and type text to work desktop apps. The company's X account, @cursor_ai, announced the feature.
The example it gave was the "iOS Simulator," which developers use to test apps for Apple's iPhone on their computers. The agent runs features it built itself in the simulator and checks whether they work as expected. Users will be able to hand off not just writing code but also testing it on screen.
Used when needed, even without a request
Users can request Computer Use through instructions, and agents will also choose it on their own if they judge it necessary for the task. It can be used both by agents running in the cloud and by agents running on machines managed by users or companies. Running it on Macs managed by a company requires explicit permission in the operating system. Cursor has not disclosed which plans are eligible or the pricing.
Also working with Grok Bot on development
Cursor is jointly developing the AI agent "Grok Bot" with SpaceXAI (formerly xAI) under SpaceX. Anthropic and OpenAI also offer computer operation as an agent feature, and its use to reduce testing work is spreading among development tools as well.
