The new builders are already inside your company

When companies decide to "get serious about AI," the first instinct is usually to look outward: hire the specialist, bring in the vendor, buy the platform. Look inward first. The people who will actually carry this are usually already on your payroll — and they're probably being quiet about it.

The revolution surfaces new, unrecognized champions inside every org. The job is to remove the social barriers that keep them from showing up.

Who they are

They're rarely who the org chart predicts. In my experience the strongest early adopters are often not the most technical people — sometimes the technical folks are the most resistant, because their identity is wrapped up in the old way of doing the work. The new builders are the curious ones. The ops person who quietly automated half their reporting. The support lead who's been drafting replies with AI for months and polishing them by hand. The bookkeeper who figured out how to reconcile statements in a tenth of the time and told no one, because they weren't sure they were allowed.

What marks them isn't credentials. It's three things: curiosity, judgment, and domain knowledge. Everything else — the tooling, the techniques, the vocabulary — is teachable. Those three are much harder to install than prompt syntax.

Why they're hiding

Because in most companies, being visibly good at AI is socially expensive. It invites side-eye from teammates, suspicion about cut corners, and awkward questions about what else could be automated. So your best people learn in private and your org learns nothing. That's the real cost of an unsafe room: not the experiments that fail, but the wins that stay hidden.

What to do about it

• Invite, don't lecture. The move is not a lesson from supposed experts down to everyone. It's an invitation up: share what you've already figured out, build a way to learn together, write it down as you go.
• Give your enthusiasts standing. Don't just tolerate them — formalize them. A small, sanctioned group with real influence over how the company explores AI, protected from normal process and politics. Think skunkworks, not committee.
• Reward the showing, not just the result. If someone surfaces a workflow they automated, the first response has to be curiosity, not an audit. The second person to come forward is watching what happens to the first.
• Restructure around what you learn. When the lessons accumulate, let them actually change how work is organized. Nothing kills an exploration culture faster than insights that go nowhere.

Before you hire your AI strategy, take a month to find out who already started building it. The answer is usually closer than you think.