All posts
AI

Codex Cloud Keeps Working While Your Laptop Sleeps, but Setup Is the Real Task

ZKO blog tile: AI

OpenAI's Codex cloud runs coding tasks in the cloud, from investigating bugs to building features, then lets you review changes and continue from the web, mobile or desktop app. Each task gets its own workspace and can keep running while your computer is asleep.

Our take: the headline is the sleeping laptop, but the real work is the first step. You choose repositories, Codex inspects them and works through setup, and you publish an environment before starting tasks. If that environment is wrong, every task inherits the mistake.

On mobile you can review work and open an existing environment, but creating a new one needs the web or desktop app. That is a sensible split. Setup deserves a proper screen, and review is what you want to do on a train.

We would treat the environment like a project template. Pin the dependencies, document the commands that must pass, and keep secrets out of it. Then a task failing means the code is wrong, not the setup.

Review stays human. The flow ends with you reading changed files and results, asking for follow-ups, and then committing or opening a pull request. Keep it that way for client work.