Building AI Champions: Who to Choose and What They Need

Key Takeaways

• AI champions are people from the business who use AI well and help others do the same. They don't need to be technical.
• Choose people the team trusts and who know the work, rather than simply the biggest technology fans.
• Without protected time, the role turns into unpaid evening work and fades away.
• Write the role down: tasks, time, who they turn to and what champions may decide.
• Leadership stays responsible: it sets the rules, protects the time and answers open questions quickly.

AI champions are colleagues from the business side who use AI well in their own work and help the people around them do the same. They aren't IT specialists. They're respected, curious members of the team who get some extra training, a few protected hours and a clear role. In a small or mid-size company, they're what keeps AI in everyday work after the first workshop.

What does an AI champion actually do?

The role is small, but its tasks are clear. A champion stays in their existing job, as an administrator, project lead or bookkeeper, and takes on a few fixed tasks alongside it:

• Uses AI in their own work and spends five minutes in the team meeting showing what worked and what didn't.
• Is the first port of call for questions like “How do I summarise a long email thread with this?”
• Collects prompts and examples that work and looks after the shared prompt library. Our guide to prompting basics shows what good prompts look like.
• Notices where rules are unclear or block a sensible use, and reports it to whoever decides on the rules.
• Spots new uses in the team's work and takes them to wherever use cases are decided.
• Helps new colleagues get started with the approved tools.

What isn't part of the role?

What champions are not matters just as much. They aren't IT support: licences, passwords, installations and permissions stay with your IT team or IT partner. They don't decide what's allowed. That's for leadership, ideally in a short AI usage policy. And they don't do everyone else's AI work. If every task gets handed to the champion, nobody else learns anything, and the champion never gets round to their own work.

Who makes a good AI champion?

Technical knowledge matters less than many people think. What counts is whether the team listens to them and whether they know the work. Look for these qualities:

• Respected for their own work, so others listen when they recommend something.
• Curious and patient, and they enjoy explaining things.
• Knows the workflows first-hand: in a fiduciary firm, someone from bookkeeping; in a trades business, the person who writes quotes or plans jobs.
• Careful: checks results and takes data rules seriously.
• Has time, or can be given it without other work piling up.

Ask for volunteers first, then approach the people who fit. Don't automatically pick the youngest person in the team or the one who tries every new tool anyway. A colleague who started out sceptical and was won over can move the team further than someone everyone already knows as the tech fan.

How much time do AI champions need?

Without protected time, the role doesn't work. It becomes extra work in the evenings, and after a few months it quietly stops. Agree a fixed, modest amount, such as an hour a week or half a day a month, and write down which other work drops away to make room. Their manager has to back that. In the early weeks, when lots of questions come in, it can be more. After that it settles down.

How do you set up the role step by step?

1. Write the role down.Describe on half a page what a champion does, how much time is set aside, who they turn to and what they may decide. That gives clarity to the champions and to the team.
2. Choose and ask.One person per team or location is a good start. In a small office, one person can cover the whole business. Ask in person and explain why you're asking them.
3. Train the champions first.Champions need a head start: deeper training on the approved tools, prompting, checking results and the data rules, before the whole team's turn comes. It's a fixed part of our team training.
4. Give them the means.Access to the approved tools, a shared place for prompts and examples, and a direct line to the person who decides on rules and tools.
5. Connect the champions.A short monthly exchange across teams: what's working, which questions come up, where things get stuck. Good ideas for new use cases surface here, which you can then prioritise with use-case scoring.
6. Make the work visible.Leadership mentions the champions' work in meetings, good examples are shared with the team and the role comes up in performance reviews. Work nobody sees soon counts as a side issue.

What do champions need from leadership?

Champions support leadership and can't stand in for it. The owner or partner has to say clearly that AI use is wanted, and within which limits. Leadership protects the time set aside, asks in meetings how things are going and decides quickly when a champion brings a question upwards. If a question sits unanswered for weeks, such as whether a particular tool may be used for client data, champions lose credibility with their colleagues. Our AI enablement programme supports exactly this part, with coaching for champions and regular office hours.

A champion without protected time is just the person who gets asked the most questions.

How can you tell whether it's working?

Satisfaction surveys tell you little. After three months and again after six, look at these points instead:

• Do the teams use the approved tools regularly, beyond the champions themselves?
• Do others use the prompt library, and are new entries being added?
• Are the questions champions receive getting more advanced over time? That's a good sign.
• Which tasks does the team do differently than three months ago, and does it save time? Our method for measuring time saved by AI shows how to capture that properly.
• How are the champions doing? Is the time enough, or is the role eating into their evenings?

Which mistakes should you avoid?

• Naming champions without telling the team. Say who they are, what they help with and what they don't, so everyone knows where to turn.
• Asking champions to enforce the rules. They can explain them, but anyone who has to police their colleagues loses the trust the role depends on.
• Ending up with the champion as the only person who uses AI well. The aim is for the whole team to improve, so keep an eye on how far the others have got.
• Keeping champions away from decisions. If their reports on unclear rules or new use cases never lead anywhere, they'll soon stop making them.
• Starting too big. One champion and one team where it works convince people more than a programme announced across the whole business.

What happens after the first year?

Give the role for a set period, such as a year, and then talk to each champion about whether they want to carry on. Rotating spreads the knowledge across more people, and former champions remain useful contacts. The page on AI in your team shows how champions fit together with training, rules and measurement. And since much of this starts with a workshop, running an AI workshop that sticks shows how to make it last.