History
Codex takes coding tasks into separate cloud environments
OpenAI introduced Codex in ChatGPT as an agent for working on multiple coding tasks side by side.

On 16 May 2025, OpenAI introduced Codex as a research preview in ChatGPT. The coding agent could work on several tasks side by side. Each task ran in its own cloud environment with the project’s code. The agent could propose changes and run tests, among other actions. The introduction describes the setup.

We see a change here in how programming work was organised. A user could have an assignment carried out separately and inspect the result later. This makes the quality of the assignment and the assessment of returned work important.

For a small or medium-sized business, a trial can start with one small change to a practice application. Supply the desired behaviour, boundaries and checks. Make sure the environment contains enough information to perform the task.
Then have a developer assess the code and test output. Use the changed application through its screen yourself. Also check that the change stays within the assignment’s purpose. Start several assignments side by side only once you can assess a separate result for each task and check the changes together.