Every Wave, Same Mistake

THE EFFECTIVENESS DOCTRINE · ON AI

What forty years of technology waves taught me about AI, and why the part that decides the outcome is the part nobody can buy.

By Doug Girvin · Transilience Partners

The first time I “surfed the web,” in 1993, there were fewer than ten websites to visit. I’d downloaded the Mosaic browser, and I sat there in a kind of quiet astonishment, clicking between a handful of pages, feeling the shape of something enormous arriving. I’d already spent the eighties with my hands on hand-me-down PCs running DOS, dialling into the world through CompuServe. So by the time the web showed up, I had learned the rhythm of these moments: the arrival, the promise, the gold rush, and then the strange, expensive disappointment that almost always follows.

I’ve now watched that rhythm play out four or five times. And I’ve come to believe the disappointment is never really about the technology. It’s about a mistake leaders make, the same mistake, every single time.

THE PATTERN, AND THE MAN WHO NAMED IT

Client-server computing was going to kill the mainframe. The web was going to end bricks-and-mortar business. “Digital transformation” was going to reinvent the enterprise, and instead became, in my experience, the most expensive exercise in paving cow paths ever devised: millions spent making yesterday’s processes run faster on shinier systems. Each wave arrived as a silver bullet, and each underdelivered for the same reason.

The reason has nothing to do with chips or bandwidth or models. When a powerful new tool arrives, leaders reach for the tool instead of rethinking the work. They take the process they already have, the one built for a world that no longer exists, and they automate it.

None of this is my discovery. In 1990, Michael Hammer wrote in the Harvard Business Review: “It is time to stop paving the cow paths.” He also said why the old processes were so hard to shift: they were “geared toward efficiency and control.” That was thirty-six years ago. We have been paving ever since.

NEARLY FORTY YEARS OF THE SAME FINDING

It isn’t a hunch, either. In 1987, three years before Hammer, the economist Robert Solow made a remark that became famous: “You can see the computer age everywhere but in the productivity statistics.” Organizations had bought the machines. The results had not arrived.

Part of the explanation came fifteen years later. In 2002, Erik Brynjolfsson, Lorin Hitt and Shinkyu Yang found that each dollar a firm invested in computers was associated with more than ten dollars of market value, depending on the model. They attributed most of it not to the hardware but to intangible assets, including the way the business was organized. The computers paid off most where the organization changed around them.

And today, BCG says the companies leading on AI follow what it calls the 10–20–70 rule: ten percent of the effort on algorithms, twenty percent on technology and data, and the remaining seventy percent on people and processes.

Four technologies, nearly four decades, one finding. So the interesting question isn’t whether it’s true. It’s why, after all this time, almost everyone still skips it.

WHY EVERYONE STILL SKIPS IT

Here is my answer, and it’s the reason I wrote this piece.

The 30% is procured. The 70% is managed.

The algorithms and the platform arrive with a budget line, a vendor, an RFP, a demo, a renewal date, and someone whose job depends on the rollout. The people-and-process work arrives with none of that. There’s no SKU, no invoice, no natural owner, and nobody whose quarter is measured on it.

It isn’t skipped because leaders disbelieve the research. Most of them have heard some version of it. It’s skipped because there is no way to buy it.

I've spent much of my working life studying how organizations are structured, and there’s a structural truth underneath this. Work without an accountable owner doesn't get done, no matter how much everyone agrees it matters. The 70% is everyone’s job, which in practice makes it nobody’s.

Put more simply, the way I now say it to clients:

The licence is the part you can procure. Adoption is the part someone has to manage.

To be fair to my own industry: the large firms do sell the 70%. They sell it in a size most organizations can't buy or absorb. That’s a problem worth its own piece, and I’ll come back to it.

AND NOW, AI

Which brings us to this wave. AI is the most capable instrument any of us has ever been handed, and the silver-bullet reflex is firing harder than I have ever seen it.

You don’t need a statistic to see the pattern. Walk into almost any organization mid-rollout. The licences are paid for. The usage dashboard looks fine. A handful of enthusiasts are getting real leverage. Everyone else has quietly gone back to doing it the old way, and nobody wants to be the one to say so in the steering committee. The tool was procured. The adoption was never managed.

I’ve also watched a darker version of the reflex take hold this time, and I want to name it plainly: the fantasy that AI is, at last, the tool that replaces the people. I’ve seen leaders in both the corporate and public sectors quietly orient their whole AI strategy around headcount. I understand the arithmetic. I think it’s a profound misread of the technology, and of where value actually comes from. It's early days, but studies are beginning to reflect the challenge for companies that have attempted to replace people with AI at scale. It's not positive. Stay tuned.

THE FEAR IS RATIONAL. SO IS THE OPPORTUNITY.

When leaders frame AI as a replacement for people, the workforce’s fear isn’t irrational. It’s an accurate reading of the room. People go quiet, use the tools shallowly and in secret, and the organization learns nothing, which guarantees the very failure the leaders then blame on the technology.

I hold two things at once here, because I’ve lived on both sides. I’ve been the founder responsible for a team’s livelihoods, and my leadership philosophy was that my job was to work for my people: to resource them, clear obstacles, and make them more capable. And I’ve been the technologist who loves what these new instruments can do. Those don’t conflict. They point to the same conclusion.

The opportunity in AI was never efficiency. Efficiency is doing the same work faster: the cow path with a faster horse. The opportunity is effectiveness: changing the work itself so the outcome gets better, and using the technology to take the drudgery off your people so they can do what only people can do.

WHAT THE FEW WHO GET IT DO

In every wave, a small minority broke the pattern and pulled away. They did the unglamorous thing. They didn’t ask, “How do we do this faster?” They asked, “If we were starting from scratch today, knowing what this can now do, how would we do this at all?”

Then they did something the others didn’t: they gave the 70% an owner and a budget line. They made it buyable inside their own organization. A named person, not a committee. One real piece of work redesigned around the answer, with a date attached. And the people brought along rather than routed around.

And they kept human judgment firmly at the helm. These instruments are astonishing, and they have no judgment of their own. They can assemble facts to the horizon and not know what a single one of them means. Meaning, intention, and what actually matters here come from people. The technology is extraordinary crew. It is never the captain.

That’s why I don’t talk about AI as a tool you point at your old processes, and I certainly don’t talk about it as a replacement for your people. I talk about it as a teammate: one you bring onto the work, give a role, direct, and remain accountable for.

THE WAVE IS ALWAYS COMING

Another wave will come after this one, and another after that. The only question that has ever mattered is whether the people meeting the wave have built the steadiness to meet it well: the judgment to know which work is worth redesigning, the discipline to keep humans at the helm, and the leadership to bring their people through the change rather than around it.

Forty years in, that’s the lesson I’d stake everything on.

The tool will always be the part you can buy. What decides the outcome is the part you have to build.

WHERE DOES YOUR ORGANIZATION STAND?

Take the free AI Readiness Index: a three-minute read on the conditions for adoption rather than the enthusiasm for it, and the first move to get unstuck. For a team ready to do the work, From Tool to Teammate runs as a self-paced course or as a private team cohort.

Sources: Michael Hammer, “Reengineering Work: Don’t Automate, Obliterate,” Harvard Business Review, July–August 1990. Robert Solow, “We’d Better Watch Out,” New York Times Book Review, 12 July 1987, p. 36. Erik Brynjolfsson, Lorin M. Hitt and Shinkyu Yang, “Intangible Assets: Computers and Organizational Capital,” Brookings Papers on Economic Activity, 2002(1), pp. 137–181. BCG, “The Leader’s Guide to Transforming with AI,” December 2024, updated July 2025.