Most companies need an AI leader well before they can justify hiring one. The decisions arrive first: what to build, what to buy, who to trust, what to refuse. They are expensive to get wrong, and they do not wait for the job advert.
We have held that seat three times: building an AI consulting practice across EMEA at a global technology leader, as Chief AI Officer at a marketing analytics company, and setting AI strategy for an upcycling business. Different sizes, same job: work out which of the AI ideas on the table are real, build those, and leave behind a team that no longer needs us.
A company at the start of its AI work hits a circular problem. It cannot hire a credible AI leader until it knows what that person would do, and it cannot work that out without someone senior enough to decide.
The gap gets filled badly. Either a vendor sets the agenda, which produces a roadmap shaped like the vendor's product, or the work is handed to whoever is technically strongest internally, which costs you your best engineer and still leaves the decisions unmade.
An interim leader gets those decisions made without adding a permanent executive to the payroll. Whether it worked shows up later, in what the organisation can still do once the seat is empty again.
There is a third way the gap gets filled: promoting someone from inside. It looks like the safe option and it carries a cost nobody names.
An internal candidate has to live with every recommendation they make. Cancelling a project means cancelling a colleague's project. Moving budget means taking it from someone they will sit opposite next Monday. Telling the board that the flagship AI initiative should stop is, for them, a career decision. For us it is simply an assessment.
The incentives differ in the other direction too. A permanent AI leader is rewarded for building a function - more headcount, more platform, more scope. We are not. We have no promotion to earn here, no department to protect, and no side in whatever disagreement predates our arrival.
That is what most clients are actually buying. Not the hours, and not only the experience: a judgement with nothing riding on it, from someone whose leaving is the point rather than the risk.
A global technology company had an EMEA AI practice that existed in name only: no methodology, no senior bench, no client relationships. Our founder built and led it from there.
Four things had to move at once: hiring senior consultants, writing down a delivery method that still worked when someone else ran it, landing enterprise accounts, and showing results fast enough for the investment to continue. None of them could wait for the others. The decision that made it scale was preferring a repeatable method over bespoke work, so delivery quality stopped depending on which person was in the room.
That is the honest argument for an interim leader over a consultancy. The company kept the capability; it did not end up depending on the person who built it.
Our founder held the Chief AI Officer seat at a company building a cross-media attribution platform. The product reconciles paid search, display, email and social with television, radio and out-of-home, into a single view of what actually drove a purchase.
In a measurement business, the AI decisions are the product decisions. Some questions are answered by multi-touch attribution, some need marketing mix modelling, and some can only be settled by running an incrementality test. Deciding which is which is the job. So is deciding when a model may simply state a number, and when it has to show its working - because a client is about to move budget on the answer.
The work delivered under that seat is written up separately in Cross-Media Budget Optimisation.
This is a circular-economy business, where the stock is whatever came back through the door and no two items are alike. That single fact decides most of the AI question. Grading, pricing and matching supply to demand are all harder here than they are for a retailer working from a fixed catalogue.
The engagement was strategy rather than build: where AI would earn its place, in what order, and which parts were not worth doing yet.
Across one of these engagements, three streams of work turned out to be solving overlapping problems, each sponsored by a different part of the business. Consolidating them took 60% out of the effort that had been going into the overlap, and what remained was pointed at outcomes with owners and dates rather than at activity.
That is the clearest example of the argument above. Everyone could see the duplication. Nobody inside was in a position to say so, because each stream belonged to somebody.
Beyond the number, we judge an engagement by what still works after we leave. Three things should be in place.
A roadmap that has been decided, not drafted. With the reasoning written down, so the next person can tell which choices were deliberate.
A delivery method someone else can run. If it only works when we are in the room, it is not a method.
The people to run it. Hired, or grown from the team you already have.
An engagement that leaves you dependent on us has failed, however well the individual projects went. We would rather write ourselves out of the role than extend it.
If you already know what you want built and have someone able to run it, you need engineers, not a leader - and we will say so. We have talked companies out of AI programmes that were not ready. Not ready usually means a data problem or an ownership problem wearing an AI costume.
The other poor fit is a company wanting the title on a slide for a funding round. The seat only works if the person in it can actually decide things.
The EMEA practice was built by our founder in a senior role at a global technology leader. The Chief AI Officer seat and the AI strategy engagement were delivered as ProDataAI.
Let's talk about what's achievable for your specific context.
Start the Conversation →