JavaOne 2026 Session

Summary

AI coding agents are powerful, but often unpredictable. They jump into implementation, miss requirements, or produce code that can't be tracked. Spec-driven development brings structure and control to this process. The method is simple: start with clear, high-level requirements, refine them into a detailed development plan, then break that plan into a task list with trackable steps. The AI agent works from these artifacts—requirements.md, plan.md, and tasks.md—instead of ad-hoc prompts. Each step becomes explicit, reviewable, and repeatable.

In this talk, we'll show how to apply spec-driven development through a practical example—defining requirements, generating a plan, creating tasks, and guiding the AI step by step. You’ll learn techniques for keeping AI workflows structured, auditable, and reliable. If AI coding feels chaotic, this session offers a framework for turning it into a dependable partner.