Asana says Codex compressed a five-year test migration into two weeks

Asana used OpenAI Codex to replace an outdated testing system in two weeks, according to OpenAI. The company estimated the work would otherwise have taken five years and said the project cost about $12,000.

The claim is notable because test infrastructure migrations are often slow, unglamorous engineering projects. They require reading old code, updating patterns, preserving behavior, and avoiding regressions across large codebases.

Codex is OpenAI’s coding agent, and this case frames it as a tool for large maintenance work rather than only writing new features. If the numbers hold in similar settings, that points to a practical use for AI coding systems: compressing repetitive but risky modernization tasks.

The report comes from OpenAI, so teams should treat it as a case study rather than a general guarantee. The useful takeaway is narrower: well-scoped migrations with clear tests may be among the strongest places to apply coding agents.