What Happens When Frontend and Backend Become One Team?
1. Background of the Integration
- Problems with the Previous Structure: Because the frontend and backend teams were separate, substantial waiting and coordination costs accumulated at role boundaries through API specification agreements, sprint scheduling, and ticket handoffs.
- Running the Experiment: The Advertiser Platform team combined four backend and four frontend engineers into one eight-person team responsible for everything from the interface to the server.
2. Key Operating Principles and Results
- Work Allocation: In every sprint, each team member took at least one ticket from the opposite discipline, encouraging gradual expansion into the other area.
- Leadership by Example: The team leader reduced fear by taking on cross-discipline work and migrations first.
- Maintaining Productivity: Even during the transition into unfamiliar technical areas, the amount of work completed in each two-week sprint stayed at its previous level.
3. Success Factors
- Leverage from AI Coding Tools: AI significantly lowered the entry barrier created by differences in frontend and backend technology stacks and programming languages.
- Team Knowledge and Documentation
- Existing domain documents, architecture diagrams, and runbooks became essential context for AI.
- Documentation and AI usage formed a reinforcing cycle.
- The Changing Role of Expertise: Engineers shifted from being “producers” who write 100% of the code themselves to “final reviewers” who judge and verify AI-generated output.
4. Costs and Limitations
- Difference Between Planning and Execution: Because AI enabled faster execution, unplanned work increased, and two-thirds of the actual work had not been included in the plan, reducing predictability.
- Reduced Synchronization Among Team Members: As individuals handled work quickly in parallel, the sense of “working together” became somewhat weaker.
- Sustainability of Individual Full-Stack Work: Managing fragmented domain knowledge and preserving depth of expertise remain future challenges.
5. Prerequisites
- A product structure where frontend and backend connect into one feature
- Mature AI tooling and established usage habits
- An existing documentation culture and prior alignment on sprint and team practices
- A small eight-person team
