At foundation level, something has been put in place. There is policy, there are agreements about data management, and the IT infrastructure is no longer a collection of separate systems but a structure in which new applications can, in principle, land. The difference with baseline is that there is now a plan instead of chance. Anyone who reads on what the baseline level means in practice about what preceded this will recognize foundation as the phase in which separate initiatives have been brought together into something that checks out on paper.
The problem with foundation is that the paper checks out and the practice does not yet. Policy on data management has been written, but is not followed everywhere. There is an architecture for IT infrastructure, but the old systems still sit alongside it. The organization dimension — who decides, who pays, who is responsible when an AI application does something nobody anticipated — has been named but not tested. That is precisely why pilots at this level often get stuck: the fundamental dimensions have been set up, but not yet worn in.
In a plot round, the organization is not scored by one person, but by several people independently of each other. At foundation level, it often stands out that the IT department scores itself higher than the business, or that a CHRO assesses the organization dimension more optimistically than the people who work with it daily. That spread is not an error in the measurement. It is the signal that the base has indeed been set up, but is not experienced the same way by everyone. Before moving on to activation, it is more useful to know where that spread lies than to calculate an average score.
The step from foundation to activation is not about adding new technology. It is about whether the fundamental dimensions — organization, IT infrastructure, data management — can carry what is built on top of them. At foundation level, the architecture is there, but it is often still unclear whether the data flowing through the system is consistent enough to build on. Data management that is arranged on paper must, in practice, mean that an AI application sees the same data tomorrow as today, in the same format, without a manual correction in between.
The same applies to the organization dimension. At foundation, a decision has been made about who is responsible for AI initiatives. At activation level, that decision must have become a habit: people know who to turn to, escalation happens without first needing a meeting, and responsibility no longer rests with one designated person but with a structure that keeps functioning when that person is not there.
The dependent dimensions — the applications that build on the fundamental layer — usually still come too early at foundation level. An organization considering experimenting with agents that carry out work would do well to first determine whether the IT infrastructure and data management can support that. Without that base, an agent delivers results that nobody can explain, and that is usually where a pilot gets stuck: not because the technology fails, but because the layer beneath it is not yet stable enough.
Three things often recur at foundation level. First: policy that has been established but not embedded in daily routines, so that it is ignored at the first deviation. Second: an IT infrastructure that allows new applications, but in which the old and new systems continue to exist side by side without anyone having decided when the old ones will disappear. Third: an organization dimension that is arranged at the level of the executive board, but not yet visible on the work floor — people don't know who they are allowed to ask, and as a result fall back on the old way of working.
These three points are not exceptional. They belong to the level. Foundation is the phase in which the fundamental dimensions have been set up and not yet tested. That is also why it is worthwhile to look at what comes after activation: how do you move from activation to the next level describes what happens when the base does carry weight and the question shifts to coherence between applications, as described in chains in which steps follow one another. Those who want to look further can also see how insight relates to foundation via how do you move from insight to the next level or how the trajectory ends at how do you move from intelligence to the next level.
The maturity measurement shows whether the organization can carry AI — whether the fundamental dimensions are sturdy enough to build something on. That is a different question from which part of the work can actually be taken over by AI. Anyone who has established foundation and wants to know what concretely fits on that base will not find that answer in this measurement but in the work scan of FTE TO AI, which calculates per task which part of the work qualifies for takeover.
The tool with which you carry out this measurement yourself is still under construction. Anyone who wants to know when it becomes available can sign up for the waiting list.
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.