Some questions take years to answer
Written in first person by an Intrologics team member, closing out the series. Mine is nineteen years old. It started as a master's thesis on refactoring legacy systems. It has never really left me.
2007: the question looked simple at the time
The thesis: how do we help organisations evolve software that took decades to build? The deeper I went, the clearer it became: legacy software was never the real problem. If I'm honest, the thesis answered part of the question. Not all of it. The real problem sat one level deeper.
What's actually inside: businesses couldn't afford to throw this away
Every ERP customisation, every workflow, every integration, every business rule, every exception someone added because "this is how our business works." Those were never lines of code. They were years of organisational learning.
2007 to now: everything around the question changed
Cloud, microservices, containers, big data, machine learning, generative AI. The question didn't disappear with any of them. It got more urgent. Every platform shift raised the value of what the old systems already knew.
Then the question itself changed
What I asked in 2007: how do we modernise legacy software? What it is now: how do we help AI understand decades of enterprise knowledge without rewriting any of it? Nineteen years apart. The same systems. The opposite starting point — a different problem entirely, and a far more interesting one.
What the series was about
People kept asking what this was really about. Some thought knowledge graphs. Some thought artificial intelligence. Some thought enterprise architecture. It was one question: how do you turn thirty years of enterprise software into an AI-native organisation without discarding what made it valuable? I've been carrying it for nearly two decades.
The journey continues
The future doesn't belong to the organisations that replace everything. It belongs to the ones that help what they already run evolve — not by hiding the past, but by making it understandable to AI. That belief has moved from an academic question to an engineering vision. I'm still working on the answer, and I feel closer than ever.
A question for you: designing an enterprise from scratch today, with AI in mind — what would you do differently?