Rakuten fixes issues twice as fast with Codex

Rakuten, a global innovation company employing 30,000 people across e-commerce, fintech, and mobile communications, has integrated OpenAI's coding agent, Codex, into its engineering stack. According to Yusuke Kaji, General Manager of AI for Business, the company relies on Codex to advance three primary operational goals: building faster, building safer, and operating smarter. In operational workflows, engineering teams use Codex alongside KQL-based monitoring to accelerate root-cause analysis and remediation, resulting in an approximate 50 percent reduction in mean time to recovery during incidents. To address potential deployment bottlenecks, Rakuten integrates Codex directly into its CI/CD pipeline. The agent conducts automated code reviews and vulnerability checks by applying internal coding principles, allowing teams to ship updates quickly while maintaining security standards. Furthermore, Rakuten leverages Codex for larger, ambiguous projects that lack fully defined specifications. For example, Codex independently built a full-stack mobile application featuring a Python and FastAPI backend alongside a Swift and SwiftUI iOS app, reducing a quarter-long development effort down to a matter of weeks. As Codex assumes greater responsibility for code generation, Rakuten is actively shifting the role of its engineers away from manual code inspection toward defining clear specifications and verifying outputs against measurable standards. This transition is being supported by hands-on workshops spanning engineering, product, and non-technical teams across the organization.