Most organizations plan AI adoption as if it’s one group learning one skill. In practice there are three — Builders, Users, and Resisters — and they need almost entirely different things from you.
Builders come from necessity, not a mandate.
I became a builder because the tool didn’t exist and no one here could make it — so I built a Copilot agent that predicts a seam going out of spec before it happens and recommends the correction. You can’t train your way to builders. They surface where a real gap meets process thinking and skill, and not everyone has that wiring.
The builder’s real job is packaging.
You don’t need everyone to build. You need a few builders who turn capability into a reusable artifact the rest can run. I handed mine forward as instructions; now a coworker uses it every day who couldn’t build it if he tried. One builder → an artifact → many users. That’s how AI actually scales — not by making everyone technical.
Then there are the resisters.
The third group won’t adopt at all — and, uncomfortably, they’re often the roles closest to replacement. That’s its own conversation, and it’s the next one.
So what, now what.
So what: “send everyone to AI training” treats three different populations as one. Now what: find your one or two builders, free them from what the org chart expects, and invest in packaging their wins so users can adopt without building.
Leave a comment