Rakuten, the Japanese e-commerce and fintech conglomerate, says it has cut its mean time to resolution (MTTR - how long it takes to fix bugs after they're detected) by 50% after deploying OpenAI's Codex coding agent across its engineering organization.
According to an OpenAI case study published today, Rakuten's teams are using Codex to automate CI/CD pipeline reviews (the automated checks code goes through before it ships to production) and to handle full-stack builds that previously took months but now wrap up in weeks. The company operates across e-commerce, banking, mobile, and streaming, so the engineering surface area where Codex gets applied is broad.
These are solid numbers, but context matters. MTTR improvements of 50% sound dramatic until you consider that the baseline matters enormously. Cutting a 4-hour average to 2 hours is meaningful. Cutting 20 minutes to 10 is nice but not transformational. OpenAI's case study doesn't specify the absolute numbers, which makes the percentage harder to evaluate.
The CI/CD automation piece is arguably more interesting than the headline stat. Code review bottlenecks are one of the biggest time sinks in enterprise engineering. If Codex can reliably flag issues in pull requests before human reviewers even look at them, that compounds across every engineer on every team.
This is the kind of enterprise case study OpenAI needs as it pushes Codex adoption beyond individual developers into organization-wide deployments. Rakuten's scale - thousands of engineers across dozens of product lines - makes it a credible reference customer, even if the details are thin on methodology.