MyRealTrip PE Shifts from 'Writing Code' to 'Design and Verification' through AI-Native Development

MyRealTrip's Product Engineer (PE) organization shared its AI-native development experience. Presenter Song Da-in, a PE, shifted the workflow so that most of the code writing is delegated to AI, while humans focus on defining 'what' and 'why' to build and verifying the results.

AI-Driven Development Case Through a Hackathon

At a hackathon last December, the team developed a flight delay information service by applying an agentic coding approach. They generated design drafts with Figma Make, connected design and code via MCP, and wrote code using Cursor. In this process, AI quickly handled design and code generation, but applying existing code conventions and team design guidelines became a bottleneck that had to be done manually by humans.

Shift in Development Perspective and Organizational Response

As the use of AI tools has become widespread, the developer's role has shifted from 'implementation' to 'design and verification.' Song Da-in, PE, primarily uses Claude Code and other tools, focusing on clarifying objectives and verifying outcomes rather than writing code directly. The introduction of MCP has reduced the burden of managing the iOS build environment and enabled scenario-based automated testing, decreasing repetitive work.

To embrace this change at the organizational level, MyRealTrip operates a PE Expert group, building processes to organize rules in a form AI can understand and share team direction. Ultimately, the team emphasizes that establishing rules with the team, rather than solving problems alone, leads to faster results.