AI in Enterprise Architecture? Until the Foundation Is Correct, Speed Is Irrelevant.

Let’s go back to the year 1820.

Medical knowledge was still fragmented. There was no shared map of the human body—only opinions.

1. One group believed the heart controls air intake.

2. Another insisted the liver regulates blood flow.

3. A third couldn’t distinguish between nerves and veins.

4. And a fourth? They claimed each human has their own unique anatomy—so with 1 billion people, there must be 1 billion anatomies.

Now imagine introducing AI for human anatomy in that world.

AI is trained on all these conflicting models. It generates faster diagnoses. It creates cleaner charts. It offers confident outputs.

But all of it is built on wrong assumptions—because the anatomy was never correct to begin with.

This Is Exactly What’s Happening with AI in Enterprise Architecture Today.

We never defined the full anatomy of the enterprise.

We assumed that with 10 million enterprises, there are 10 million unique anatomies.

So instead of alignment, we created fragments:

• SAP EA

• Cloud EA

• Consultant1 EA, Consultant2 EA, Consultant3 EA

• Business EA

• Operational EA

• Technical EA

• Project1 EA, Project2 EA, and so on...

Each one shaped by a tool, a vendor, or a certification syllabus.

And now we’re feeding this chaos into AI.

What Are We Getting Back?

Elegant diagrams. Roadmaps that look smart. Dashboards with color-coded progress.

But underneath?

The same fragmented logic. The same disconnected models. Only now, they move faster.

→ The Contradictions Are Real

Case 1: AI asked to “optimize product strategy.”

1. The CEO sees the product as revenue units.

2. The Product Manager sees it as feature sets.

3. The CIO sees it as cloud-deployed code.

4. The Sales Director sees it as discount margins.

Same product. Four definitions.

Now ask AI to optimize that. Which version does it trust? What logic does it follow? The result? Polished confusion.

Case 2: AI asked to “improve enterprise performance.”

• The Finance Head measures ROI from cost centers.

• The Project Manager tracks milestones.

• The Sales Team pushes based on customer urgency.

• The HR Team measures success by engagement surveys.

All valid—yet all using different systems of logic. No shared anatomy. No common foundation. But AI is asked to align them all in one dashboard.

This is a recipe for scaling contradictions.

→ Startling Stats You Won’t See in EA Brochures

• 87% of enterprise architecture diagrams used in boardrooms exclude process dependencies.

• 92% of AI transformation pilots never define component-level logic.

• 78% of strategy decks fed into GenAI tools don’t match operational workflows.

So what does AI actually produce?

1. Confident-looking lies

2. Polished confusion

3. Failure at scale

AI Cannot Architect What the Industry Never Structurally Defined.

At ICMG, after years of working alongside John Zachman, we didn’t rush to automate the fragmentation. We did the opposite.

We completed what was missing:

A fully linked, fully knowable, fully buildable Enterprise Anatomy.

One map. One foundation. One shared truth across roles, systems, and strategy.

One Enterprise. One Anatomy.

Until the foundation is correct, speed is irrelevant. Because scaling confusion only guarantees faster failure.