Agentic Coding, Intentional Design

Use Domain Driven Design as guardrails to steer AI coding agents away from "AI slop" toward intentional, evolvable designs.

AI coding agents are exceptionally good at generating boilerplate code, are somewhat effective for business logic, and often produce "AI slop."

Domain-Driven Design was created to manage complexity. Using explicit patterns and a shared language, business logic stays intentional and expressive. Agentic tools default to framework-driven development and anemic models, not Domain-Driven Design.

In this session, we explore how AI coding assistants can amplify good domain design, allowing developers and architects to focus on what matters most: the business logic itself. We begin by live coding ("vibe coding") a solution with minimal guidance, intentionally surfacing common failure modes such as blurred bounded contexts and misplaced business logic. We then introduce a series of Domain-Driven Design "guardrails" that reshape the coding agent's behavior: enforcing clear bounded contexts, leveraging ubiquitous language, encapsulating business logic within expressive models, and maintaining proper separation of concerns. These guardrails become the primary mechanism for steering AI-generated code toward meaningful behavior and evolvable designs.

Attendees will leave with practical, experience-backed techniques applicable to agentic tools such as Claude Code, Copilot, Cursor, and Warp, as well as tool-agnostic patterns for integrating AI coding assistants into a DDD-centric workflow—retaining human ownership of the domain model while leveraging AI for speed and efficiency.

← Back to all talks