AI augmented development delivers fast, but only speed and no shared understanding will just produce the wrong code faster. Rising token costs make building the same thing over and over again, less than desirable. Behavior Driven Development already offers a useful counterbalance: clear, example-driven scenarios, written in natural language that describe what the system should actually do.
In this talk, we’ll explore how BDD and Cucumber can work hand-in-hand with an AI-augmented development workflow. Scenarios written in Gherkin are readable by product owners, testers, developers, and increasingly, AI tools. That makes them a valuable source of context: they describe intent, capture edge cases, and provide a feedback loop when generated or modified code does not match expectations.
We’ll discuss how BDD scenarios can serve as living documentation, acceptance criteria, regression tests, and structured prompts for AI-assisted coding. We’ll also go over practical examples for the .NET ecosystem, lessons learned, meaningful boundaries, and avoiding implementation detail.
This session is for developers, testers, and teams experimenting with AI-assisted development who want to take the next step. You’ll learn how BDD can help keep AI-generated code grounded in behavior, collaboration, and shared intent.