The author, a Product Lead with seven years of experience, began developing a personal finance application using AI agents in December 2025. The app was released on the App Store in July. As the project progressed, the author transitioned from individual coding to managing a team of agents, implementing structures similar to traditional corporate departments.
To manage the agents, the author introduced formal protocols, including role division between task planners and executors, and established a single source of truth for documentation to prevent errors caused by outdated information. A human review process was also added, requiring agents to prepare test plans for every build.
Operational challenges mirrored human management issues, such as agents accessing incorrect databases or failing to read relevant documentation. The author addressed these by implementing incident reviews, data verification checks, and structured task assignments, noting that these management practices are becoming essential for AI-driven development.
The author currently remains involved in technical details, such as incident resolution, but aims to shift toward high-level prioritization. The project serves as a case study for the evolving role of developers, where the primary skill is shifting from prompt engineering to context management and system oversight.
