In 1982, business owners asked me whether they should buy a personal computer.
It was the wrong question. Not because the answer was obvious — it wasn’t, and the ones who hesitated had reasonable grounds. It was the wrong question because the answer, either way, told them almost nothing about whether the thing would help.
I started in business technology that year and spent the next seventeen years running a firm that advised companies through the arrival of the PC, the networked office, and the early commercial internet. Since then, I’ve worked through e-commerce, mobile, and cloud — as a technology executive, a business owner, and an advisor sitting across the table from other owners.
Four transitions. Each one produced a set of companies that came out meaningfully stronger and another that came out weaker, and in almost every case, the gap between them was not explained by who bought what or when.
Here is what it was explained by.
The earliest adopters didn’t win
This surprised me for about a decade, and then it stopped surprising me.
Plenty of companies bought personal computers early. A large share of them used the machines to do exactly what they had been doing on paper, only now with a machine involved. The paper workflow stayed. The approvals stayed. The person who retyped everything stayed, and now retyped it into a terminal. Costs went up, and nothing else moved.
The same thing happened again with websites in 1998, with apps in 2011, and with cloud migrations that lifted a badly designed system onto someone else’s servers without changing what made it bad.
Adopting the technology while keeping the workflow is the most reliable way to spend money on a transition and get nothing in return. It happened in every wave. It is happening right now.
The companies that won could answer a narrow question
Not “should we get computers?” Something much smaller:
What specifically does this change about how we do the thing we do?
The firms that got value from the PC were the ones that could say: our month-end close takes eleven days, four of them are recalculations, and this eliminates those four. That’s a claim you can check. It tells you what to buy, who needs training, what the workflow becomes, and whether it worked.
The firms that got nothing had an answer that sounded like a strategy and functioned like a wish.
The cautious ones weren’t wrong — the ones waiting for certainty were
I want to be careful here, because “move fast” is bad advice sold as courage.
Some of the best-run companies I worked with moved late and did fine. They weren’t paralyzed. They were watching for something specific — a competitor’s capability they could see, a cost they could no longer carry, a customer expectation that had visibly shifted. When the trigger arrived, they moved decisively because they had been thinking about it the whole time.
The ones who got hurt were waiting for the technology to stop changing so they could evaluate it properly. That moment never arrives in any transition. It isn’t arriving in this one.
So what’s actually different about AI
Two things, and only two, in my judgment.
The capability changes faster than the evaluation cycle. With the PC, you could take eighteen months to assess a purchase, and the thing you eventually bought was roughly the thing you assessed. That’s no longer true. Any evaluation process built to produce certainty will produce a stale answer instead.
Adoption is already happening inside your company, whether you approved it or not. This one is new. Nobody smuggled a mainframe into the building. But in most companies I look at, a meaningful share of staff are already using AI tools for real work — drafting client correspondence, summarizing documents, analyzing data — and in most cases nobody has told them what’s permitted. The technology arrived before the decision did. That reverses the sequence every prior transition followed, and it means the first question isn’t “should we adopt this?” It’s “what is already happening here, and is it safe?”
Everything else about this transition rhymes with the last four.
Three questions worth asking this quarter
Where in this company does work get done twice? Not “where could we use AI.” The waste comes first; the tool comes second. Every transition I’ve watched rewarded companies that started from a specific inefficiency and punished companies that started from a product demo.
Who here is already using it, and on what? Ask directly, without consequence attached, or you’ll get a useless answer. What comes back is usually both more widespread and more casual than leadership expects — and it tells you where the real opportunities are, because your staff has already found them.
What would we have to believe for this to be worth the investment? State it as a testable claim before you spend. If you can’t state it, you’re not ready to spend. If you can, you’ve also just defined how you’ll know whether it worked — which is the step almost everyone skips.
The part that hasn’t changed since 1982
Every one of these transitions arrived wrapped in language suggesting that the technology itself was the point, and that the companies that bought the most of it the soonest would come out ahead.
That has not been true once.
What has been true in all four is that the businesses that came out ahead were the ones whose leadership understood their own operation well enough to say precisely what the new thing changed about it. The technology was the easy part. The judgment was the hard part, and it still is.
AI is the most consequential of these transitions. It is not the first one that has required business owners to make important decisions without complete information — and the ones who did that well before are doing it well now.
Ken Nangle has worked in business technology since 1982. He advises owners and senior leaders of established companies in Sonoma, Napa, and Marin Counties on the transition to AI — whole-company reviews, prioritized plans, applied training, and rapid application development.
This article first appeared on LinkedIn.