AI and Architecture: Future or Folly?
From: AEA Webinars - AI and Architecture: Future or Folly?
I recently listened to a webinar by Chris Armstrong, President of Armstrong Process Group, regarding the intersection of AI and Enterprise Architecture (EA). While the session focused on how AI fits into established frameworks like TOGAF, Chris offered a lot of wisdom between the lines on what it actually means to be a “good” architect. This is exactly the kind of perspective I needed when I first started in this field.
The Core Concept: Abstraction
A central theme was the principle of abstraction—the ability to represent the essential characteristics of a system while de-emphasising the irrelevant. Chris highlighted that what makes an abstraction “good” is inherently subjective and varies by domain.
“An abstraction represents the essential characteristics of an entity that distinguish it from all other kinds of entities… What makes an abstraction ‘good’ is a matter of perspective.”
He made a clear distinction between Architecture Building Blocks (ABBs), which focus on the “what,” and Solution Building Blocks (SBBs), which focus on the “how.”
Where AI Fits (and Where It Does Not)
The webinar captured the current pain points of EA work: keeping models current, correlating content, and the mundane struggle of wrestling with tool user experience. Chris suggested that AI could help raise the level of abstraction, allowing architects to focus on “true” ABBs while AI handles the solution-level assets.
However, he identified several “unique capabilities” that should remain in human hands:
- Subjectivity: Determining what is “architecturally significant.”
- Collaboration: Building consensus with stakeholders.
- Innovation: Applying best practices in a specific, human context.
“Architects should not surrender their unique contribution and creativity… There are critical activities that are best suited to remain in human hands.”
My Take
While Chris captured the current struggles of the profession perfectly, I think he is being very modest about the impact of generative AI. I believe it will have a much larger effect than he suggests, automating a significant portion of the workflow and making architecture data far more reliable and up to date.
That said, he was spot on regarding why certain parts of the job must remain human. Strategy, deep understanding, and human interaction are not things AI can replicate. His examples of why we build models—to build consensus and explore alternatives—remind us that architecture is as much about people as it is about diagrams.
