Key Takeaways
"We already have AI champions."
I hear this in almost every conversation now, and it is new. For a long time I made the case to partners that change needs champions: people inside the organization who carry a shift further than any program team or leadership deck ever will. For fifteen years that was a pitch I had to fight for. Today it is a given.
That shift should feel like a win. Mostly it feels like a warning. Because when you ask the second question, "what are those champions actually supposed to achieve?", the room usually goes quiet. Push a little further and it turns out "champions" almost always means the power users. Nobody has decided what they are for, whether they understand what is expected of them, or whether they are pointed at a problem worth solving.
First, a word about the word. The term "champion" gets used in very different ways. In Six Sigma it means the executive sponsor. Elsewhere it is interchangeable with change agent, ambassador, or simply the resident expert. Here it means something specific: an employee selected and trained to help the organization change, on top of their regular job. An AI champion might help a teammate use an approved tool, sense-check its output, or work out which tasks it is actually suited to. Their influence comes from being a trusted peer, not from a position on the org chart.
This distinction matters, and getting it wrong is the first way organizations waste the whole idea. If your "AI champions" are in fact your sponsors, your technical specialists, or an undefined mix of both, you have a group with the right title and nobody responsible for helping colleagues change how they work.
There is a reason this matters more with AI than with anything we have worked on before. AI rolls out at the speed of licenses. Whether the tools arrive at all is a top-down decision, but the moment they are there, everyone can take part. You have a seat, and that is it, you are in. A company can put a tool in front of thousands of people long before any of them know how to use it well. Each person still has to work out where it helps in their own job, what they can safely put into it, and how far to trust what comes back.
Then add the thing that is specific to AI: no other change topic carries this much fear. People do not necessarily trust the leader who might cut their role. They believe the colleague two desks over who tried the thing and reported back that it was fine. Peer credibility is doing more work here than in any transformation I have seen.
The barrier to participation has never been this low. Which means the leverage of champions has never been this high. Done well, champions close the gap between access and everyday use: a colleague who understands your work shows you a relevant example, sits with you while you try it, and talks through what went wrong. That is what a general training session rarely gives you.
High leverage, high trust, low barrier. And then we waste it. Three ways.
In AI adoption there are two kinds of champions, and they are not the same people.
Power users are pulled by the technology itself. They do not need to be from IT, they just have an affinity for it. You never had to explain the tool to them, because they had already taught themselves. They arrive with use cases: "I saved five hours this week, and here is how." They are valuable precisely because they experiment, find the useful applications, and work out where the tools fall short.
Social influencers are the people others watch. They like change, or at least they accept that things have to move, and they are good at making it safe to be a beginner in public. People trust them, ask them questions, and admit confusion to them without feeling behind.
Some people are both. Most are not. The mistake is assuming that enthusiasm for the technology automatically makes someone good at bringing less confident colleagues along. Those are two different asks, and right now you need both: the power users pushing the frontier and generating the lighthouse use cases, the influencers saying it is better to be on the train than to watch it leave.
Get this wrong and it shows up in small, costly ways. Someone fascinated by the tool stands up in the team weekly and enthuses about saving fifteen hours, and does not necessarily think about the maths everyone else is mentally doing about what happens when enough people save fifteen hours. Curiosity is what makes a power user valuable. It is not the same skill as reassurance.
Safety is the clearest parallel. The EHS director is not the safety champion. That is a job. The safety champion is the shift lead who runs the toolbox talk, who calls out the shortcut her colleagues take when nobody is watching, and who has the standing to stop the line. Nobody hired her for that. She was chosen, trained, given time, and trusted, on top of the work she already had. And notice what nobody argued about: that it cost something. No serious company ran its safety turnaround by asking for volunteers and hoping for the best. Hold that thought.
Choose champions for their ability to learn, explain, and earn trust, then pair anyone who lacks the complementary half with someone who has it. And make the responsibilities explicit: who they support, what they help with, and when to hand a problem up.
A lot of organizations reach for champions because they assume it costs nothing. Champions can absolutely be a highly cost-effective instrument. That is different from free.
Helping colleagues takes time. So does staying useful: a champion has to keep testing tools, developing their own skills, and keeping up with a field that moves every week, or they fall behind their own colleagues and lose the credibility that made them worth choosing. My usual recommendation is ten to fifteen percent of their working time, and that time has to come out of their existing workload. Their manager needs to agree what they will stop doing.
The other half of this is prioritization, and it is worth being blunt. The conversation I have now is rarely "is this the right thing to do." People agree that it is. The conversation is "it is right, and I still cannot get it through." Everything is being stress-tested. If you have ten must-haves and can fund six, and champions sit at number seven, it simply does not happen. And often it is not even money, it is attention: there is already so much change on the pile that nobody can add another item to it.
So the question is not whether champions are a good idea. It is whether they are aimed at a problem important enough to survive the cut. "Increase AI adoption" is not that problem; it tells no one which work should improve or how to judge progress. Give them something specific instead. A sales team might ask its champions to help colleagues use an approved tool to prepare proposal drafts, then track preparation time, rework, and whether quality holds. The champions get a clear focus, and managers can see whether the support is actually working.
That focus also exposes what champions cannot fix. If faster drafting changes little because every proposal waits a week for legal sign-off, the fix now needs the teams who own that process. When an AI initiative meets the wider organization, it meets existing incentives and approval rules, and a champion can name the delay but cannot reorder another department's priorities on their own.
Point champions at your two biggest adoption problems, define what changes if it works, and make it measurable. Otherwise it is a band-aid, and everyone can feel that it is.
What you actually want is a reservoir: a group across the organization you can activate again and again, because you already know the change is not going to stop.
Sustainability champions, then digital champions, then AI champions, each network built from scratch and discreetly abandoned, is an expensive way to keep starting over. Some of what a champion knows will always be specific to the topic; an experienced AI user does not automatically know how to help a team cut its environmental impact. But the part that transfers is the valuable part: the ability to listen, explain, involve colleagues, and carry a concern to management. That is wider change capability, and it is worth keeping. It is fine for people to have a favourite topic. It is not fine to rebuild the network every time the topic changes. Update the membership and the training as needs shift; keep the relationships and the skill.
There is a related pattern I see in exactly the moments when it hurts most. In periods of high uncertainty, champions start to be experienced as a threat to the hierarchy, and the response is to shrink them back to pure subject-matter expertise, which conveniently makes them harmless and useless at the same time.
That conflict is a false one. Healthy change logic has both: clear decisions, made and carried visibly by the hierarchy, and a strong network in which champions are the nodes that put those decisions into practice and report back what is working or getting in the way. For that feedback to be worth anything, managers have to be willing to act on it. Remove either side and the change stalls.
Be honest about the mandate.
A champion cannot take away your fear about what your boss does with the time you free up. They cannot decide how the company will use that time. Those are top-down calls, and pointing a champion at them is unfair to the champion and unconvincing to everyone else. They need responsible AI leadership and direct communication from the people actually making the decisions.
What a champion can do is help you make the jump: get over the threshold, use the thing already sitting on your desktop, get more out of it, and share what worked. That is a narrow mandate, and a powerful one, because with AI the threshold is lower than in any change topic we have worked on.
And there is a version of this I keep coming back to. Whatever happens to roles and job profiles over the next few years, the most responsible thing you can do as a company is give as many of your people as possible a real chance to get good at AI. Training is part of that. So is having someone nearby who understands the work and can help when you get stuck. Whatever they do next, and wherever they do it, being bad at this will make it hard.
Before you appoint more champions, ask:
Almost everybody has champions now. Very few are getting the return. The difference is not effort. It is clarity about who they are, what they are for, and what you are willing to invest.
We help organizations define the role of their champions, select and train the right people, and connect their work to measurable adoption goals.
Talk to us about turning AI access into everyday use.
