Yogiyo Pilots 'Product Engineer' Model to Resolve Productivity Stagnation Despite AI Tool Adoption

Yogiyo adopted AI coding tools such as Claude Code, Codex, and Cursor to improve development speed and code quality, but overall organizational productivity did not significantly improve. While individual developer work speed increased, the total lead time from planning to deployment did not decrease.

Eliminating Bottlenecks by Dissolving Role Boundaries

The cause of the productivity stagnation lay not in the development phase but in the hands-off structure between roles. Loss of context during handoffs between planners and developers, wait times, and rework costs offset the speed improvements provided by AI tools. To address this, Yogiyo defined the Product Engineer role, where one person handles everything from planning to implementation, adopting an approach that completely dissolves role boundaries.

Pilot Operation on Live Services

To ensure reliable results, Yogiyo selected actual Yogiyo services rather than side projects as the pilot target. Participants from mobile, Frontend, Backend, and planning domains crossed their areas of expertise to perform each other's tasks, experimenting with mixing roles. Through this process, they are simultaneously diagnosing the viability of the Product Engineer model and identifying actual bottlenecks in the product development process.