The development
What happened
Holiday Extras says Codex, OpenAI's AI coding agent, helped its team reach a working mobile-app rebuild in two weeks. A 12 November OpenAI customer story reports better booking conversion and fewer errors afterward, but compares different periods rather than establishing that Codex alone caused the changes.
Key facts
Holiday Extras reports a 32.5% rise in main-app open-to-book conversion after the rebuild. OpenAI says this compares different periods; the story supplies no controlled comparison establishing Codex's causal contribution.
The reported main-app error rate fell from 19% to 9%: a 10-percentage-point drop, not a 10% relative reduction. The source does not give the measurement windows or underlying counts.
The team says it reached a working rebuild within two weeks, making progress it estimates previously would have taken more than six months. That is a company-reported estimate, not a measured end-to-end production release comparison.
Design and business expertise shaped prototypes, while engineers handled reusable packages, architecture, APIs, security and development workflows. One agent-prepared customer-requested change was approved by a named executive before submission.
The main app reportedly moved from an annual major release to weekly releases in the first month after the rebuild. A separate Countdown app also accelerated iteration; its cadence should not be mistaken for the main app's.
The context
Why it matters
TideReach assessment: the concrete AI mechanism is software development with Codex, coupled with engineers retaining responsibility for production foundations. The case suggests that customer-facing staff can turn domain knowledge into prototypes sooner. The reported conversion and error figures describe one rebuilt product across different periods; without comparable time windows, controls or counts, they cannot establish a general productivity multiplier or prove that AI alone improved the business outcomes.