When to hire a Head of AI instead of us

The budget for a Head of AI is approved, or close to it. Maybe the search is already open. And underneath it there is a doubt you have not said in the meeting: you are not certain there is enough work to keep that person busy, or that a hire is even the right shape for the problem.

Start with what the seat costs, because most of the advice skips it. Heidrick & Struggles surveyed 151 US data and AI executives in 2025. The median total cash was $300,000, and the median equity on top of that was another $200,000. So the middle of the market is roughly half a million a year, and the average sits higher because a handful of packages clear a million and pull it up. Nearly every one of these roles carries a bonus and most carry annual equity, so it is not a number you pay once. It is a standing line that grows.

Now read the advice about whether to spend it. Almost all of it is written by firms like ours, and almost all of it arrives at the same place: do not hire, engage a consultancy, preferably the one whose logo is at the top of the page. That should make you wary. It makes us wary. So we will do the opposite. We will tell you when to make the hire and not call us.

Hire the Head of AI when AI is the product, not a layer laid over it. If the model is the thing your customers are paying for, and it will keep being that thing, you need it owned inside the building by someone whose entire job is to make it better every quarter. That is not a project with an end date. It is a permanent function, and permanent functions get an owner on the payroll.

Hire when the work runs for years, not for a quarter. A Head of AI with one hard problem and then a long tail of maintenance is an expensive way to feel current. If the roadmap holds a single serious AI build and then steady state, you do not need a standing executive. You need the thing built and your team trained to run it, which is a different purchase with a different price.

Hire the person you already have. Plenty of companies are sitting on a Head of Data or a staff engineer who sees the problem clearly and is waiting for someone to hand them the mandate. In IBM's 2026 survey of CEOs, the share of companies reporting a chief AI officer went from 26% to 76% in a year. A jump that fast is mostly retitling, not net new hiring, which tells you how many firms already had their person and finally named them. If yours exists, you do not need a $500,000 hire and you do not need us. You need to give them the title and the budget, then get out of the way. We have talked a company out of a search on exactly this. It cost us the engagement.

The hire is the wrong move when you cannot yet name the work the person would own. That is the doubt from the first paragraph, and it is the one to watch. A search does not wait for that answer. It fills the gap with a job description and trusts that a good hire will find the work once they arrive. Four months and a large salary later, you have a talented person and nothing shipped, and a cost you cannot yet point at a result.

There is a shape between a permanent hire and nothing, and it is the one most mid-market companies need first. A senior operator embedded part-time, owning the roadmap and the vendor decisions, until you know whether the work justifies a full seat. That is our Fractional AI Leadership engagement, and part of the deal is that we tell you when you have outgrown it and should hire.

If you cannot say which of these you are, that is the real first purchase, and it is cheap next to the salary. One month, fixed fee, an operator maps where AI actually fits your work and comes back with the answer: hire, build, embed a fractional lead, or do nothing yet. That is the whole point of the AI Readiness Assessment. It exists so you do not spend half a million a year answering a question a month can settle.

We can afford to send you off to make the hire because we do not bill by the hour and we do not staff a bench that needs feeding. Our founder ran products at PopCap, NaturalMotion and Wooga, and has hired into the seat this essay is about. We take two engagements a quarter, which means turning down a poorly-shaped deal costs us nothing we wanted.

So before the search opens, answer one question. Can you name, in a sentence, the work this person owns eighteen months from now? If you can, hire, and hire well. If you cannot, do not close the gap with a salary. That is the conversation we would start with, and some of the time it ends with us telling you to go hire.