There's a scarcity problem in AI that doesn't get discussed enough. The people who genuinely understand how to apply AI well to a business are in short supply, expensive, and mostly working for large companies or building their own products. For an SME, hiring that expertise is difficult and retaining it is harder.
The common response is to bring in a consultant. Sometimes that helps. But the businesses we've seen make durable progress with AI usually have something the consultant can't provide: someone on the inside who owns it.
The distinction matters, because a consultant and an internal champion solve different problems, and confusing the two is a common way AI initiatives stall.
What the champion actually does
The internal champion isn't necessarily your most technical person. They're the person who understands the business well enough to see where AI fits, and who cares enough to push it through the friction of actually deploying it.
Day to day, the role looks like this. They notice the repetitive, time-consuming work across the business that might be a good AI candidate — because they understand what people actually spend their time on. They have enough technical literacy to tell the difference between a use case that's tractable and one that sounds good but won't work. They own the unglamorous middle of a deployment: defining what "good output" looks like, setting up the review step, noticing when the system drifts. And they're the person other people go to when they want to try something but don't know how to start.
None of this requires a data science background. It requires business context, reasonable technical literacy, and the temperament to push something through rather than admire it from a distance.
Why an external consultant can't be the champion
A consultant can be genuinely useful — for a specific assessment, a technical build, a sharp opinion when you're stuck. What they can't do is carry the context.
AI deployments succeed or fail on details that are specific to your business: what your data actually looks like, where the edge cases are, what "good" means in your context, which corners can be cut and which can't. That knowledge accumulates through being there day after day. A consultant parachutes in, builds something against a snapshot of your business, and leaves. The deployment then has to survive without the person who understood why it was built the way it was.
The other issue is incentives. A consultant is motivated to deliver a project and move on. A champion is motivated to make the thing actually work over time, because they're still there when it doesn't.
This isn't an argument against ever using external help. It's an argument that external help works best when there's someone internal who owns the outcome and can direct it — and stalls when it's brought in to substitute for that person.
How to back a champion
If you have someone who could be this person, the most useful things you can do are unglamorous.
Give them time. The champion role is real work, and treating it as something to do on top of a full existing role is the most common way it fails. Even a day a week, protected, is more useful than enthusiastic encouragement and no capacity.
Give them permission to start small and occasionally fail. Early AI projects don't all work. A champion who's punished for a project that didn't pan out will stop proposing them, which is the opposite of what you want.
Give them air cover. Deploying AI changes how people work, and that generates friction. The champion needs visible backing from leadership when they hit resistance, or the resistance wins.
Finding the champion you might already have
In a lot of businesses, this person already exists — they're just not labelled. They're the one who automated something in a spreadsheet that they weren't asked to automate. The one who's already using AI tools in their own work and quietly getting more done. The one who keeps suggesting things could be done better.
That instinct — toward making things work better, with the technical literacy to actually do it — is the thing to look for. It matters more than seniority or job title.
The honest version of the talent scarcity problem is this: you probably can't hire your way to AI capability as an SME, and you may not need to. The capability you need is more often about backing the right internal person than acquiring an external expert. Finding and supporting that person is a more reliable path than the consultant most businesses reach for first.
