How to Design System Architecture Diagrams with DDD in the AI Era

Architects often need to present domain‑level or platform‑level architecture diagrams for reporting, but such high‑level views typically lack the granularity required to direct development teams.

In the AI era, large language models can quickly generate code that meets quality and standards, making it feasible to connect architectural intent directly to implementation through well‑crafted prompts.

The proposed approach is to create a module‑level functional architecture diagram that incorporates Domain‑Driven Design (DDD) principles. This diagram sits beneath the platform‑level architecture (which lists system modules, their positioning, and inter‑module relationships) and maps to concrete application packages.

The platform‑level diagram provides an overview of modules but cannot guide a frontline development group. By adding a DDD‑based functional diagram, architects link each module to an "application layer"—coarse‑grained packages such as a real‑time transaction unit or a batch‑processing unit—and a "domain layer" defined according to the architect’s domain understanding.

This structure forces architects to deeply study the domain to produce meaningful abstractions, while developers can create application packages following the diagram, using frameworks like COLA that support DDD‑styled package organization.

For example, the book "Programmer's Underlying Thinking" illustrates a division that balances technical and domain dimensions, as shown in the following image:

Within each application package, outer packages correspond to the domains identified, and inner packages are organized by functionality.

If an organization enables large‑model capabilities, architects can directly translate the functional diagram into standardized prompts, allowing AI to generate the corresponding application code automatically.