What happens when you build software with AI in front of an audience?

In this live session, we’ll ideate with the room and turn an idea into a working solution using AI-assisted development practices.

Getting AI to produce code is the easy part. The engineering challenge is getting it to build the right thing, to work effectively inside a codebase, verify its own work, and keep the solution reliable as it evolves.

We’ll tackle those problems using techniques such as Spec-Driven Development, context engineering, harness engineering, AI-assisted Test-Driven Development, and agentic development. We’ll use specifications to capture intent and constraints, engineer the context available to the AI, structure how agents plan and execute work, and use tests and automated verification to close the feedback loop.

Rather than treating these as separate methodologies, we’ll see how they fit together into a practical AI development workflow, from idea to specs and context to implementation and testing.

This is not a polished demo. We’ll build something live, make decisions as we go, inspect what the AI produces, and deal with the problems that emerge.

You will leave with practical patterns for structuring AI-assisted development: how to give AI enough context and constraints to work effectively, how to create feedback loops that catch mistakes, and how to move from “AI that writes code” toward AI that can reliably participate in software engineering.