First know where the value sits. Then pick tools.
At Google I built the AI adoption for a department of 1,500 people. Your AI lead, without putting one on the payroll.
Book a call →Thirty minutes, no obligation. You hang up with at least one thing you can use.
1,500
people, AI adoption built at Google
54%
adoption within six months, in a department with roughly €30 billion in revenue
95%
of investments in generative AI return nothing
AssessmentWhere your organisation actually stands. Measured, not guessed.
DirectionOne application that removes a real bottleneck, and what deliberately waits.
AdoptionYour people on board, so it keeps working once I am gone.
IndependentNo licences, no vendor, no agency stacking up hours.
The real bottleneck
The problem is rarely the technology
You can build the most elegant automated workflow there is. If rubbish goes in, rubbish comes out. Nobody makes diamonds out of dirt.
The preparation almost always comes too late. An idea gets discussed, a pilot gets built, and only then does it turn out that the underlying data is a mess. On invented data everything works. Support inside the business only appears in a real situation, and that is exactly where it trips.
The foundation under an AI application is not the model but the data beneath it. If that is not in order, you build something that looks convincing in a demonstration and falls over in practice.
95% of investments in generative AI return nothing. MIT Media Lab and Project NANDA, 2025.
The approach
AI adoption starts with the problem, not with the technology
A fractional AI officer owns the direction and stays accountable for the outcome.
- 01Assessment. I map where your organisation stands, by having your people place themselves on a scale. You learn where you stand, they learn what to watch for.
- 02Direction. One application that removes a real bottleneck, and a clear list of what is deliberately not on the table yet.
- 03Adoption. An internal rollout is a product launch aimed at your own people. Test first with those who can handle disruption, only then go wide.
- 04Handover. Your own people become the ones who carry it. Lasting dependency on me is a failure, not a business model.
If adoption lags, that is my responsibility. It means the application does not solve the problem, or it was brought to people badly.
Prioritisation
Every opportunity gets three scores
The model used inside Google to weigh AI opportunities, applied to your organisation. It removes most of the list of ideas before anything gets built.
Value
What does it return
In money, in time, or in work that no longer has to be done. An opportunity without a number is an opinion.
Usability
Can people work with it
Does it fit the way your people work now, or does it demand a change nobody is going to sustain.
Feasibility
Can it be built
With the data, the systems and the people you have today. Not with what might be there in two years.
Also possible
Having doubts about your Google Ads?
A report every month, and still no answer to whether your money is being spent well. At Google I worked with the largest advertisers in the Netherlands and I know where the waste sits.
Next step
A good conversation is the best first step
Thirty minutes in which we go through your situation.