Article
Should you hire a head of AI?
“We need someone for AI.” The sentence comes up in many executive committees. It’s right: the decisions to make are numerous, technical and binding. But it doesn’t say who, or in what form.
There are usually three options. None is right in absolute terms.
The decisions waiting for someone
Before choosing an option, you have to measure what needs deciding. For a company that wants to move beyond experimentation, the list often looks like this:
- Which use cases to prioritise, and which to drop?
- Which model providers: an American player, a European one, open models hosted in-house?
- Which architecture: where do the models run, how do they access the data, how are permissions managed?
- Which usage policy for employees, who already use AI tools, declared or not?
- What preparation for the AI Act and security requirements?
- Build or buy: which skills to bring in-house, which providers to choose, and how to evaluate their work?
These decisions are structural. They commit budgets, contracts and architecture choices that are hard to undo. They can’t wait for a recruitment to succeed.
Option 1: hire full-time
It’s the obvious solution, and sometimes the right one. If AI is at the heart of your product or business model, you’ll need a dedicated leader, present every day.
But recruitment takes time, experienced profiles are rare and expensive, and the role is hard to define until the strategy is set. The risk is hiring too early a profile that doesn’t match the real needs, or too late, after the structural decisions have been made without them.
Option 2: hand the topic to IT
The IT department knows the information system, the security constraints and the suppliers. Giving it AI makes sense. But it’s often already stretched by keeping things running, and generative AI needs specific skills: evaluating model quality, RAG and agent architectures, controlling inference costs.
This option works well when IT can rely on an experienced perspective for the first decisions, while it builds up its skills.
Option 3: a fractional technical leader
A fractional CTO or Chief AI Officer works a few days a month, with real responsibility for technical decisions, not just an advisory role. They set the roadmap, settle architecture choices, frame the providers and report to the management.
This formula has two advantages. It gives immediate access to experience that would take months to recruit. And it leaves time to understand the real need before hiring: the day a full-time role is justified, the profile is clear, and the fractional CTO helps find it.
It also has its limits: it doesn’t replace a team, and it needs an internal counterpart available to carry the day-to-day.
How to choose
A few questions help decide:
- Is AI your product, or a lever on your business? If it’s your product, aim to hire eventually.
- Have the structural decisions already been made? If not, an experienced perspective available quickly beats a hire in six months.
- Do you have a technical team in place? A fractional leader needs internal relays to be effective.
- Can you describe the role? If the job description stays vague, it’s often a sign that the strategy needs setting first.
In many cases, the right answer is a sequence: a fractional technical leader to lay the foundations, then a prepared hire, and an organised handover. The exit is part of the contract.
If you’re asking yourself the question, a scoping call helps you see clearly in 45 minutes.