AI coding agents can now generate Spring Boot APIs incredibly quickly. But speed creates a new problem: how do we know the generated implementation actually matches the API we designed?

A service can compile, start successfully, and even pass generated tests while still drifting from its contract through incorrect fields, enum values, status codes, validation rules, or response behaviour.

In this session, we’ll explore a spec-driven development workflow where the API contract remains the source of truth. We’ll start with a reviewed API specification, use it to guide an AI agent while building the Spring Boot implementation, and continuously validate the running service against the expected contract.

We’ll also look at how contract failures can be fed back into the coding agent to create an iterative repair loop until the implementation behaves as expected.