Geek Out Time: From Ontology to AI Agents -Determinism Matters in High-Compliance Industries

In my earlier geek out, “Why Enterprise AI Quietly Depends on Ontology”, we explored a simple but easily overlooked idea: enterprise AI doesn’t fail because models aren’t smart enough. It fails because reality isn’t modeled explicitly enough.

AI agents are getting better at reasoning, planning, and explaining themselves. In demos, they feel confident, fluent, and decisive. In regulated environments, that confidence is exactly the problem.

Industries like finance, insurance, healthcare, and payments don’t just care about answers. They care about where an answer came from, which rulebook applied, what version of policy was in force, and what the system did when it wasn’t sure. A correct outcome without context is still a failed system. Very quickly, a few non-negotiable requirements show up.

Decisions must be tied to explicit sources, not just “model reasoning”. If a conclusion cannot cite which rules or guidelines it relied on, it cannot survive review.

This is an excerpt — the full article continues on Medium.