Developers have been using AI code completion tools for several years now, but AI coding agents such as Cline and Windsurf are really coming into their own. Everyone has seen games and apps being built in minutes. These tools are incredibly powerful for greenfield projects and prototyping, but they also come with some massive frustrations as codebases become more complex.

While the promise of AI-powered development is enticing, the reality often includes challenges like context limitations, dependency management issues, and technical debt accumulation. As projects grow beyond simple demos, developers need structured approaches to effectively collaborate with AI agents and maintain code quality. Without proper guardrails, AI-generated solutions can quickly become unwieldy and difficult to maintain.

Agile methodologies provide a framework that addresses many of these challenges. By breaking down complex problems into AI-manageable components, implementing iterative refinement cycles, and maintaining human oversight through regular code reviews, teams can harness the speed of AI while avoiding its pitfalls. This session will demonstrate these principles in action by building a real application with AI assistance, highlighting practical techniques that work in production environments.