The question comes up often: should we recruit a "head of AI"? For an SME, the answer is no if you mean a post, and yes if you mean a role. A company of ten to two hundred people does not have enough work for a full-time AI post, but it does need someone in the team to drive day-to-day use. That is the AI lead. Here is whom to choose, what they do and what they do not do.

Whom to choose

Someone from the existing team, chosen on three criteria:

  • Comfortable with digital tools. They do not need to be in IT.
  • An AI user. Someone who already uses it for their own tasks and can see what works.
  • Available and known. Colleagues know who they are and how to reach them. An AI lead nobody can find is no use.

It is often the person who started using AI before everyone else. You appoint them, set aside about an hour a week for the role, and tell the team.

What the AI lead does

Four tasks, about an hour a week:

  1. Answering questions. "Can I put this document into the tool?", "How do I get a shorter answer?". This is the heart of the role.
  2. Checking that the rules hold. Is the sheet on permitted data being followed? Has a new tool appeared? Have the accounts of people who left been closed?
  3. Reviewing progress with management every month. Fifteen minutes: what is working, what is stuck, what needs deciding.
  4. Being alerted if there is an incident. If protected data goes into the tool or a mistake reaches a client, they are the first to be told and they set the agreed plan in motion.

It is a coordination role, not an expert one. The AI lead does not need to know more about AI than we do; they need to know more than their colleagues about how the team uses it.

What the AI lead does not do

  • They do not decide the budget. Management decides, using the figures they report.
  • They do not develop anything. Agents and connections to your software are built by specialists, in-house or external.
  • They do not run courses. The initial training is the 4-hour AI Workshop; the AI lead answers questions afterwards. Our guide to training your team describes this format.
  • They do not carry the responsibility alone. Management sets the framework (which data, which tools, who approves); the AI lead puts it into practice.

Why not a dedicated post

A dedicated post costs a salary, a box on the organisation chart and an expectation: AI becomes "their" subject. In practice, that person oversees use that happens elsewhere, in the teams, and the gap widens between what they decide and what colleagues actually do. The AI lead, by contrast, is part of that use and sees what is getting stuck the same day. A post may make sense later, in a company of several hundred people running several projects at once.

Why not hand the role to an outside provider

Because use happens in your team, on your files. A provider does not answer everyday questions and does not see what gets stuck between two meetings. It does have a part to play, though: supporting the AI lead. Under AI Adoption Support, we prepare the settings, answer the difficult questions, hold the monthly review with them and update the rules when a tool changes.

What changes once use is established

After three to six months, the AI lead answers fewer questions: the team knows the rules and manages on its own. Their time shifts to new uses and checks. That is the sign that adoption has taken hold. Conversely, an AI lead who is still answering the same basic questions after six months points to a problem with the rules or the tool, not a need to lighten their load.