The Forcing Function

When one company in your industry starts using AI seriously — not as an experiment, as an operating principle — the first thing that happens is small. A workflow that's been queued behind a single overworked person gets unstuck. The dispatcher's tribal knowledge gets captured in a system that works while she's on vacation. Quotes that took three days now take three hours. The compliance backlog that's been a constant low-grade emergency starts working itself down.

That alone is a meaningful edge. But the edge isn't what matters. What matters is what happens next.

Once that workflow is unstuck, the next bottleneck becomes visible. The planner upstream is now the slow point — so they get a tool too. Then procurement approval. Then the reporting cycle. Then the hiring process. Each unlock exposes the next, and the company's clock speeds up at every step.

The company that didn't start isn't falling behind linearly. It's staying the same speed while its competitor compounds. After twelve months the gap isn't 20%. It's structural.

This is where the math gets uncomfortable.

When a competitor quotes in hours and you quote in days, you don't just lose deals — you lose the ones you'd have won on relationship, because relationships don't outlast a 5x speed difference for long. Their audit documentation stays current because a system keeps it that way; yours takes three weeks to assemble, and the inspector sees the difference before they ask a question. They hire and onboard in two weeks. You take ten. The people you wanted are already working somewhere else.

None of those are productivity gains. They're existence gains. The competitor isn't smarter than you. They're operating on a different clock.

And it's not only competitors setting the pace. When regulators start seeing what AI-assisted documentation looks like from the facilities that adopted early, that becomes the baseline they expect from everyone else. It won't take long before someone asks whether AI-assisted safety management is inherently safer technology — and in regulated industries, that's not a philosophical question.

Elon Musk named this dynamic in a conversation with Peter Diamandis on Moonshots. AI in its current state — today, before any next-generation models ship — can already handle roughly half of all white-collar work. But companies don't move just because something becomes possible. They keep doing what they've always done. The only thing that forces change is when one company starts beating another by using AI more aggressively — what Musk called "a forcing function for increased use of AI." On Nikhil Kamath's podcast he put it more viscerally: a supersonic tsunami. Not a slow tide. Not a cycle. A wave that's already moving.

The reasonable objection is that you've heard this before. ERP was going to fix everything. IoT was going to fix everything. Digital transformation was going to fix everything. None of them did, at least not the way the consultants promised. Skepticism is earned.

What makes this different isn't the slogan — it's the deployment cost. ERP took three years and twenty million dollars to roll out. AI augmentation, applied to one workflow at a time, takes weeks and runs on data you already have. You don't bet the company on it. You prove it on one workflow, then the next, then the next. The risk profile is unrecognizable.

In every company there are a handful of people carrying institutional knowledge that lives only in their heads. The dispatcher who knows which routes work. The compliance manager who knows which regulator cares about which detail. The maintenance planner who knows which machine is about to fail. They are irreplaceable, overworked and undocumented. They are also exactly the people AI can take the most weight off — capturing their judgment in systems that don't forget, freeing them to spend time on the calls that still require a human. Safety-critical decisions. Novel situations. Long-built customer and regulator relationships. Mentorship. The judgment calls where there's no rule that fits.

The companies adopting AI carefully are doing it to support those people. The companies that don't will lose those people first — not to AI, but to competitors who learned how to support them properly.

Most companies that don't adopt won't go under. They'll just shrink, slowly, in ways that look like normal industry attrition. But some will lose contracts they can't replace, watch their best people leave for faster competitors and discover at year three that the gap has compounded past the point where catching up is mathematically possible. Which group your company ends up in won't be obvious from the outside until it's too late to choose.

The forcing function doesn't ask permission. It arrives whether you've prepared for it or not.

The question isn't whether your company should be adopting AI. It's whether you do it now, on your own terms — or later, after a competitor has used it to take the choice away.

That gap is not closing.