Hiring an AI lead: profile, job description and alternatives for an SME
In most SMEs, you should not hire an AI lead from outside but appoint one from your current team, with 10 to 20% of their time set aside and a written mandate. An external hire only makes sense if you already run several AI uses in production and need full-time coordination. This article gives you the profile to look for, a usable job description and the criteria to choose between hiring, appointing or getting outside support.
The AI lead is part of a broader approach, described in our guide to AI in business for SMEs. Here we only deal with the role itself: who, with what mandate, and how to find them.
Hire, appoint or outsource: the first decision
Most SMEs with fewer than 100 employees do not need a dedicated position. The workload does not justify it, and an external hire takes several months to deliver, because they first have to understand your business.
Three situations call for three different answers:
- You are starting out (a few tool subscriptions, scattered uses): appoint a volunteer employee, train them, and set their time.
- You have established uses across several departments and sensitive data involved: a half-time AI lead, or a hybrid role (digital project manager, IT manager) that adds AI to its scope.
- You are deploying custom tools or automation at scale: a dedicated hire can be defended, or outside support for the launch phase, as our comparison of build in-house or buy in explains.
One regulatory landmark avoids a false debate: Regulation (EU) 2024/1689 (the AI Act) does not require you to appoint an AI lead. The European Commission's FAQ on AI literacy states that no specific governance structure is mandated to comply with Article 4. We covered this in the article appointing an AI officer: what the AI Act says. The role answers an organisational need, not a legal obligation.
What an AI lead does in an SME
An AI lead has four duties, and the first one is not technical: knowing who uses what. Tools used without anyone's approval are common (this is what shadow AI covers), and the inventory starts there.
- Map uses and tools. Who uses which assistant, with which data, for which task. It is the basis for everything, including a possible AI audit.
- Set the rules. Carry the company AI policy: what can be entered into a tool, what never is, who approves a new tool.
- Build team skills. Not by lecturing, but by being the person people ask, and by organising the right training (see Article 4 of the AI Act and training).
- Track and report. A few simple indicators reported to management: active uses, measured time savings, incidents.
What is not part of the role: building tools, negotiating vendor contracts alone, or acting as a lawyer. The AI lead coordinates and escalates; they replace neither management, nor the data protection officer if you have one, nor your IT provider.
The profile to look for: credibility before technique
The right profile is someone teams already listen to, not necessarily someone who can code. Basic technique is learned in a few weeks. Internal credibility cannot be decreed.
Qualities to look for, in order of importance:
- a good grasp of the company's trades and their real constraints;
- teaching ability, especially explaining without jargon;
- rigour with data (what can be shared and what cannot);
- practical curiosity: the person tests tools personally and can describe their limits, for example hallucinations or drifting answers;
- direct access to management, otherwise the role stays decorative.
On the technical side, three skills are enough at the start: prompt engineering, understanding what a language model does (and does not do), and the basics of GDPR applied to AI tools.
Internal profiles that work most often: administrative or HR manager, project manager, marketing or quality manager. Profiles that work less well: someone already at 100% workload, or a purely technical profile with no link to the business.
Sample job description (to adapt)
An effective job description fits on one page and sets 90-day deliverables, not a list of qualities. Here is a template you can reuse.
| Item | Suggested content |
|---|---|
| Title | AI lead (cross-functional mission) or Head of AI and digital uses |
| Reporting line | General management or operations management |
| Time allocated | 10 to 20% at the start, to be reviewed at 6 months |
| Duties | Inventory of uses, running the AI policy, supporting teams, tracking indicators |
| 30-day deliverables | List of tools and current uses, first risks identified |
| 60-day deliverables | AI charter or policy approved by management, data entry rules |
| 90-day deliverables | Training plan, 3 indicators tracked, first review at the management committee |
| Skills | Business knowledge, teaching ability, GDPR basics, hands-on use of AI assistants |
| Budget | Training for the AI lead, tools if any, time deducted from the original role |
Two useful points. First, the dedicated time must be taken off their other duties, otherwise the mission is added to an already full workload and fades within three months. Second, if you do hire (permanent contract or work-study), pay depends on region and experience level: rely on recent market data rather than a figure frozen in an article.
Where to find this profile and how to assess it
The most reliable pool is internal: the challenge is spotting who already uses AI seriously. A two-question survey is enough to bring volunteers forward: who uses an AI assistant at least once a week, and for what.
For an external hire, three exercises beat a classic interview:
- A 45-minute case: ask the candidate to propose three AI uses on one of your real processes, with the data involved and the risks.
- A scenario: an employee pasted a client file into a consumer assistant, what do you do?
- A request for proof: a concrete example where the candidate set up a use and what they measured.
Be wary of profiles that promise huge gains without mentioning constraints. A good AI lead can tell you where AI is useless.
Common mistakes to avoid
The first mistake is to appoint someone without giving them time or a mandate. The others:
- confusing an AI lead with a technical expert, and hiring a profile too far from the business;
- aiming for a full-time position too early, before there are uses to manage (see AI deployment mistakes in SMEs);
- forgetting change management: an AI lead without visible management backing is not heard;
- not training the AI lead first: they need a common base before training others;
- setting no indicators, which makes the role impossible to defend at the next budget decision.
What budget to plan
For an internal appointment, the main cost is the time taken from other duties, plus initial training. For a 30-person SME that appoints an AI lead at 15% of their time, that is roughly three quarters of a day per week: this time has a payroll cost you can estimate simply from the person's loaded salary.
For training, funding schemes exist depending on your OPCO (French skills fund) and your size: see our article on funding AI training with your OPCO. To put this cost in perspective with the rest of your project, our analysis of the cost of AI for SMEs gives orders of magnitude.
Frequently asked questions
Is it mandatory to appoint an AI lead? No. Regulation (EU) 2024/1689 does not require this role, and the European Commission states that no specific governance structure is mandated for Article 4. It remains useful to organise uses and meet the AI literacy obligation.
Does an AI lead need to know how to code? No. The useful skills are business knowledge, teaching ability, rigour with data and good hands-on practice with AI assistants. An overly technical profile, far from the field, is often less effective.
How much time should it take? Plan for 10 to 20% of a full-time role at the start in an SME, to be reviewed after six months depending on the number of uses. This is an indicative benchmark and depends on your size and sensitive data.
How is it different from a data protection officer? The DPO is a function framed by GDPR in specific cases. The AI lead is an internal organisational function with no legal status. The two can cooperate, or even be held by the same person if the required skills and independence allow it, to be checked case by case.
Can we use an outside provider instead of hiring? Yes, especially for the launch phase: audit of uses, AI policy, initial training. The provider prepares the ground, then an internal AI lead takes over day to day.
Who should the AI lead report to? Reporting to general management or operations gives the most weight. Reporting to an isolated department limits the reach of the role.
How do you measure an AI lead's effectiveness? With three to five indicators: number of approved tools, share of trained employees, active uses, measured time savings on specific tasks, data-related incidents.
How GrowthPerf supports you
GrowthPerf is a Qualiopi-certified training organisation. We train future AI leads in SMEs and associations, and help them set up their first framework: inventory of uses, AI policy, training plan. Depending on your starting point, the operational AI training for SMEs or AI awareness for businesses is the right foundation. To understand your company's context before choosing the role, you can also start from an AI maturity assessment.
If you are unsure whether to appoint, hire or get support, a first conversation helps set the decision criteria. Contact us to discuss it, and find the full framework in our guide to AI in business for SMEs.