Foundations First: Four Pillars Every Enterprise Needs Before AI Twenty years of building data platforms teaches you something that nobody puts in the AI hype cycle: the algorithm is the easy part. I have spent my career on the layer underneath. The data pipelines, system architecture, and governance infrastructure connect raw enterprise data to the models that eventually use it. It is unglamorous work compared to training a model or fine-tuning a transformer. But it is the work that decides whether AI actually makes it to production or quietly stalls in a staging environment somewhere. In December 2025, I gave a talk on “ What Needs To Be In Place Before AI ” for the IEEE Richmond Computer Society, you can find here YouTube , and the response told me this is something a lot of people are wrestling with. So here is the written version: four things that need to be in place before you touch a model, drawn from real systems and real failures.…