Foundation is the level at which an organization no longer starts from zero. Choices have been made, there is policy, there is a team or a role dedicated to it. It is the step after baseline, in which AI happened mostly incidentally and on the initiative of individuals. At foundation level, something has been established. The question is no longer whether the organization is engaged with it, but how evenly this is spread across the seven dimensions that the maturity measurement distinguishes.
The most common form of foundation is uneven. Organization and policy are often the first to be in place: an AI officer has been appointed, there are guidelines on use, perhaps a first governance document. But IT infrastructure and data management regularly lag behind. Systems are not set up to support AI applications, data is scattered, quality and accessibility are not assured. This is not an exception; it is the pattern in which the foundational dimensions lag behind the dependent ones.
This shows up concretely: a pilot that has been approved on paper gets stuck because the required data is not available or not reliable. A policy document exists, but no one in the organization can explain it without pulling it up. A person responsible has been appointed, but that person has no mandate to make decisions about systems or budgets. Foundation looks more orderly on paper than it works in practice.
At foundation level, the spread in the plot round is often at its widest. A CIO rates the IT infrastructure low because he knows what still needs to happen in terms of system integration. A CHRO rates organization and policy high because there is now at least a point of contact. Both judgments are correct from their own position. The measurement makes that difference visible rather than averaging it out, and that is exactly where foundation calls for a conversation: which dimension is ahead, which still needs to catch up, and is that difference in sequence known to everyone who has a say in it.
The reason foundation presents itself this way lies in the nature of the dimensions themselves. Organization and policy can be set up relatively quickly: a document, a role, a meeting structure. IT infrastructure and data management require more time, because they depend on existing systems that were not built for AI. Those who at foundation level have invested mainly in policy and organization get an impression of progress without the foundational layer having grown along with it. That difference only becomes visible once a pilot wants to move beyond the test setup, and it turns out the data is not ready for it.
This is also why foundation is an intermediate step and not an endpoint. The level at which this resolves is called activation, where the foundational and dependent dimensions move closer together. The levels after that, insight and intelligence, build on a foundation that at foundation level is not yet equal everywhere.
The step from foundation to the next level does not call for more policy. There is already policy. It calls for bringing the lagging dimensions up to the same level, particularly IT infrastructure and data management, so that the dependent dimensions — including AI use and the degree to which employees are familiar with it — have something to build on. That means systems need to be put in order before new applications are unleashed on them, and data quality needs to become a topic with an owner, not an assumption.
The route from here to the next level is described at how you move from foundation to the next level. Those wondering what the step before this one looks like will find it at how you move from baseline to the next level; those who are already past foundation can read on at the page about the step from activation to the next level.
The maturity measurement shows whether the organization can carry AI: whether the foundational layers are strong enough to build something on. It says nothing about which work qualifies for this or which part of a role AI could take over. That is a different question, with a different instrument. Once the foundational dimensions are in order — or while work on them is underway — the work scan from FTE TO AI can map out which part of the work, task by task, qualifies for takeover. The two measurements complement each other: one shows whether the foundation is there, the other shows what can be done on that foundation.
Vraag maar wat er moet staan voordat AI in uw organisatie kan landen.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.