Cua
Give every agent a cloud desktop
Cua is a cloud desktop platform designed specifically for AI agents, providing sandboxed environments across Linux, Windows, macOS, and Android. It offers agent runtimes, evaluation benchmarks, and training data tooling for computer-use agents built with models like Claude, Gemini, and others. Developers can define custom environments, snapshot them, and fork into parallel runs for testing and reinforcement learning pipelines.
Cua provides hot-start cloud sandboxes with a display, browser, and root access, along with SDKs for agent runtimes, screenshot/click/shell control, and an eval benchmark to test agents on real desktop tasks.
AI engineers and developers building computer-use agents
Background.
- Status
- launched
- Business model
- freemium
Similar projects.
Editorial take on the space this project sits in — momentum signals, adjacent moves, our call on whether the wedge is real. Get pinged when we publish a new read or when the landscape shifts.
Have a take on this space?
Tell us what you’d build differently, where you think the incumbents miss, or what we’ve gotten wrong about this project. Comments + reactions are coming soon.