Kennisbank

How do you move from activation to the next level

What activation is

At the activation level, AI is no longer on the drawing board. There are pilots, there are tools in use, experience has been built up with what does and doesn't work. This is the level at which most organizations stay longer than planned. Not because nothing is happening, but because what is happening remains fragmented. One team works with agents that carry out work, another team experiments separately with a language model for text, and no one has a complete picture of where AI in the organization actually stands.

Activation is the level of the isolated application. There is proof that it can work. There is not yet a structure that makes that proof repeatable outside the place where it originated.

How you recognize activation

A few signals often come together at this level. There is at least one application that demonstrably delivers something, but the question of why it isn't being deployed more widely remains unanswered. There is budget for experiments, but no fixed process to determine which experiment gets a follow-up and which doesn't. Different departments each have their own supplier, their own approach, their own definition of success. And the people running the pilots are often not the same people who should be deciding on scaling up.

That last point is perhaps the clearest characteristic. At activation, the knowledge of what works lies with those carrying it out. The decision about what follows lies higher up. Between those two there is often no fixed connection.

In the plot round of the maturity measurement, this is the level at which the spread between scores is greatest. An IT manager sees progress because systems are running. An HR director sees a standstill because there is no policy on who may work with which tool. Both observations are correct. They simply describe a different dimension.

What needs to be in place to move forward

The step from activation to insight does not call for more pilots. There are usually already enough of those. The step calls for something that activation itself does not produce: a way to look across applications, compare them with each other, and make a choice based on that.

That starts with the fundamental dimensions. If the organization has no shared picture of who decides on what, every application remains an island. If data management is not in order, what works in one pilot cannot simply be transferred to another department, because the underlying data looks different there. And if IT infrastructure differs per team, merging separate initiatives costs more than starting over would.

This is the reason the order of the seven dimensions is not arbitrary. An organization can score high on a dependent dimension — for example, a team that is skilled with a specific tool — while the fundamental dimensions beneath it are not yet carrying that weight. That difference is exactly what is often overlooked at activation: the progress that is visible does not always rest on a foundation that allows for further growth.

Insight is the level at which those fundamental dimensions do start to count. Not because suddenly more AI is being used, but because there is then a way to see which application can be repeated elsewhere and which remains tied to the circumstances of one team. What that requires in practice, from measurability to a shared overview across applications, is described on the page about how you move from insight to the next level.

Why activation does not resolve itself

There is an assumption that more time or more budget will automatically resolve activation. That is not always true. An organization can remain stuck at activation for years, with ever new pilots that leave the same questions unanswered. The problem then is not a lack of initiative. It is a lack of a structure that connects initiatives with each other.

That structure looks different at activation than at the levels before it. Anyone wanting to know what the foundation looked like before there were experiments will find that on the pages about what the baseline level means in practice and what the foundation level means in practice. That comparison makes visible that the leap to activation mainly requires courage — trying something out — while the leap after that requires something else: overview.

Overview does not come from a single pilot. It comes from laying out everything that is running side by side, and from a way to see where that comes together in chains in which steps follow one another rather than continuing to exist separately, side by side.

Where this runs alongside the work scan

The maturity measurement shows whether the organization as a whole is ready to carry AI: whether the fundamental dimensions are in place beneath the isolated applications that already exist. That question is different from the question of which part of the work itself is suitable for takeover. Anyone who, at activation, is mainly stuck on the question of which tasks within a team or process are suitable for transfer to AI, and to what extent, will find that answer in the work scan of FTE TO AI. It calculates per task which part of the work AI can take over, offering the additional picture alongside the readiness that is mapped out here.