Unlocking Agentic Engineering with Spec-Driven Development

AI-assisted coding often breaks down when we ask LLMs to do too much at once. Large prompts, sprawling requirements, and overloaded context windows can lead to missed instructions, hallucinated features, and unreliable output. This session introduces Spec-Driven Development (SDD) as a practical way to make agentic engineering work in the real world by giving coding agents a focused, structured context rather than a single giant wall of instructions. You’ll learn what Spec-Driven Development looks like in practice, why it improves code quality, and how OpenSpec makes it easier to adopt in both new and existing projects. We’ll cover the core OpenSpec workflow, ways to customize it for your team, and common FAQs about getting started. Whether you’re experimenting with AI coding tools or trying to make them usable at scale, this talk will give you a concrete framework for turning AI from an unreliable assistant into a more effective engineering partner.