Rakuten Uses OpenAI's Codex to Cut Error Fix Time by 50%

Rakuten, the Japanese e-commerce and fintech giant, has been integrating OpenAI’s Codex coding agent into its engineering workflows to boost both speed and safety. Over the past year, the company has seen a 50% reduction in mean time to recovery (MTTR) for incidents and can potentially compress project timelines from quarters to weeks. Yusuke Kaji, Rakuten’s AI General Manager, has focused on three priorities: building faster by using Codex for root-cause analysis, building safer through automated code reviews and vulnerability checks in CI/CD, and operating smarter by using Codex to tackle ambiguous projects from spec to implementation. This isn’t about raw code generation. It’s about shipping safely at high velocity. Why it matters: Rakuten shows that agentic AI isn’t just for cutting development time; it’s a proven tool for increasing operational resilience and security in large-scale production environments.