AI coding tools are getting faster. The bottleneck is no longer the code, it is the clarity of intent. When intent lives only in a chat session, it disappears the moment the session ends. This workshop teaches you how to capture intent in a form that survives: in the repository, readable by humans and AI alike, and solid enough to ship to production.

What you will learn:
We start with the fundamentals of working effectively with AI coding agents: context engineering, the Research-Plan-Implement loop, and why structure matters for output quality. From there, the workshop shifts into spec-driven development: writing proposals, specs, and task lists that give AI agents something meaningful to work from, and that give you and your team something meaningful to review. We use OpenSpec as the practical framework throughout, covering its lifecycle, artifact structure, and how it fits into a real team workflow.
Throughout the day, theory and hands-on labs alternate. You will work through the workflow individually and as a team, covering collaboration patterns, review practices, and how to keep AI-generated output at a production quality standard across the whole team, whether you are starting from scratch or working in an existing codebase.

What you will walk away with:
A spec-driven workflow you can apply next week, both solo and with your team. A project structure you set up yourself using OpenSpec. A clear mental model for where AI-assisted development breaks down and how structured intent addresses that.

Who this is for:
Developers and tech leads already using AI coding tools who want to move beyond ad-hoc prompting toward something repeatable, team-ready, and production-grade. Basic familiarity with Git is assumed. No prior experience with spec-driven development required.

Format: A mix of theory, hands-on labs, and team exercises. Bring your laptop.