Agentic Coding, Intentional Design

AI coding agents are exceptionally good at producing working code. Give a coding agent a feature request and a Quarkus project, and it will produce a working Java app. However, working code doesn’t necessarily follow good design, and agents often produce code that ignores architectural boundaries and scatters business logic across an application’s layers.

Domain-Driven Design exists to manage complexity through explicit models and a shared language, isolating business logic and clearly implementing business rules. DDD also provides critical structure and constraints for AI-generated code. In this session, we explore how AI coding assistants can implement good design, allowing developers and architects to focus on the modeling decisions and business rules that define the system.

We’re going to build the same feature twice: once with only the requirements and once with repository Skills tailored for DDD and the Quarkus ecosystem. We will dive into how bounded contexts, business rules, and separation of concerns can be encoded into persistent agentic guidance so that our architecture survives multiple conversations.

Attendees will leave with practical techniques for turning architecture into persistent agent guidance, structuring Skills for Java and Quarkus applications, and evaluating whether AI-generated code preserves the architecture they intended.