AI initiatives are usually described in terms of models, but they succeed or fail on data. A capable model fed poor, poorly understood or poorly governed data will produce confident answers that no one should act on. The model is rarely the constraint. The state of the data underneath it almost always is.
This is uncomfortable because data foundations are expensive, slow to build and invisible in a demo. It is far easier to fund the visible thing on top. But the gap between a promising pilot and a dependable production system is very often a data gap, and it does not close on its own once the pilot is over.
Quality, lineage and access control
Quality is the first hurdle and the most obvious once it bites. If the underlying records are incomplete, inconsistent or out of date, the AI inherits every flaw and amplifies it at scale. Knowing where the data comes from matters just as much. Lineage tells you how a figure was produced and whether it can be trusted for the decision at hand, and it is what lets you explain, later, why the system said what it said.
Access control is the discipline that too many programmes leave until last. Who can see which data, under what conditions, and how that is enforced is not a compliance afterthought. It determines whether you can use the data safely at all, particularly where it is personal, commercially sensitive or regulated. Get it wrong and the choice becomes stark: expose something you should not, or lock the data away from the very initiatives that need it.
AI does not fix bad data; it acts on it faster and more convincingly than anything before it.
Treating data quality, lineage and access control as foundations rather than fixes changes the economics of everything built on top. Each new use case starts from trustworthy, well-understood, properly governed data instead of paying to rediscover and clean the same problems again.
It is unglamorous work, and it rarely makes the announcement. But it is the difference between AI that stays in the lab and AI that people are willing to rely on. The foundation is not a phase to get through. It is the thing that decides whether the rest was worth doing.