A few years ago, we rolled out a digital data-capture tool called Hash Data.
The idea made sense. Instead of collecting information on paper, people could use tablets and digital forms in real time. The company paid for the subscription. Leadership saw the upside. The use cases were clear. There was real excitement behind it.
On paper, it looked like progress.
In practice, it never gained real traction.
People kept going back to paper. Some of that was technical. Forms had to be built and maintained. Internet access was inconsistent. Only some key people were trained. But the deeper issue was bigger than any one obstacle:
We introduced a digital tool into a system still built to run on paper.
The software was not the problem.
The system around it was not ready.
Looking back, that is the part that matters most. We treated the rollout like the solution. We assumed that once the better option existed, people would naturally work differently.
They did not.
Not because they were lazy.
Not because they were incapable.
Not because the idea had no value.
They were working inside the reality they had, not the one leadership hoped for.
I think about that now because I am watching the same mistake happen again with AI.
Organizations everywhere are buying copilots, licenses, platforms, and pilots. The spending is real. The urgency is real. The confidence is real too. If we get the right tool in place, value will follow.
But in most organizations, it does not.
Not because the tools are weak.
Because the people and workflows around them were never prepared.
That gap is bigger than most leaders admit. McKinsey reported in January 2025 that almost all companies are investing in AI, but only 1 percent say they are at maturity, meaning AI is fully integrated into workflows and driving substantial business outcomes.
That is the part most AI strategy misses.
The gap nobody budgets for
Over the last ten years, across nursing, manufacturing, and HR, I have learned something that keeps proving itself:
The hardest problems are almost never technology problems.
They are people-and-system problems dressed up in technology language.
When a leader tells me, “We rolled out an AI tool and adoption is low,” I do not start by asking which tool they chose. I ask a different set of questions.
What did people hear about why this mattered?
Who helped them understand what would change in their work and what would not?
Who translated the tool into the reality of an ordinary Tuesday?
Who sat with the people closest to the work long enough to see where the friction would actually be?
Most of the time, the answer is some version of no one.
This is the gap leaders keep underestimating: the distance between buying a tool and building the conditions for people to use it well.
I call it the preparation gap.
That gap is where pilots stall, new tools become side projects, and expensive software turns into one more thing people work around.
The tools keep getting better.
The humans are not getting more prepared.
When it comes to AI adoption, the system leaders keep forgetting to build is the human system.
The context.
The trust.
The translation.
The repeated sensemaking that helps someone answer a simple question:
What does this change about how I actually do my work?
Without that, even a good tool becomes one more burden people are expected to absorb on their own.
That is not adoption.
That is abandonment with better branding.
Why training is not preparation
When I say we need to prepare the humans, I do not mean we need more training videos.
Training matters. It is just not enough.
Training teaches people how to use the tool.
Preparation teaches people how to think differently about the work because the tool now exists.
That distinction matters.
Training says: here is how to use AI to draft the report.
Preparation says: here is what changes when drafting gets faster. Here is what still depends on your judgment. Here is where context matters more, not less. Here is how your role shifts when the routine part speeds up.
One is skill transfer.
The other is thinking transfer.
Most organizations do the first and skip the second.
Not because the second is optional.
Because the second is harder.
It takes time, observation, and real familiarity with the work. It requires someone who understands the operation well enough to explain what is actually changing, for whom, and where the risks are.
When that work is skipped, people tend to do one of two things.
They ignore the tool.
Or they comply without understanding it.
Ignoring is easy to spot. Usage stays low. People keep doing the work the old way.
Compliance without understanding is harder to spot and more dangerous because it looks better on paper. The numbers say the tool is being used. The dashboard looks active. But the quality of the work does not improve. Friction just moves somewhere else.
Both outcomes are expensive.
Both are avoidable.
What the workaround is really telling you
Long before AI became the topic in every leadership meeting, I learned that good systems do not work just because they are installed.
I learned it in care environments, where the documented process only tells part of the story.
I learned it in operations, where the real workflow is often different from the official one.
And I learned it again in manufacturing, where rollout after rollout followed the same pattern: leadership buys, implementation starts, some people get trained, and six months later the real operation is still running through paper forms, side notes, spreadsheets, verbal handoffs, and workarounds.
That pattern is not random.
It reveals something important.
People do not adopt tools in the abstract.
They adopt them inside workflows, habits, constraints, relationships, and pressure.
If the system around the tool does not support the new behavior, people will go back to whatever keeps the work moving.
That is not a character flaw.
That is system logic.
The operator who writes the numbers down on paper first is not necessarily being difficult. The supervisor who keeps a shadow tracker is not automatically resisting change. The frontline team that quietly returns to the old process is often telling you something true:
the new system does not fit the work yet.
That is why I keep coming back to what I call Operational Thinking.
Operational Thinking is the discipline of seeing work as a system before trying to change it.
You do not start with the tool.
You start with the work.
You look at how the work actually happens, where the handoffs are, where the trust lives, where the friction sits, where judgment matters, where routine dominates, and where change will break if it is introduced carelessly.
Only then do you decide what technology should do.
Most AI strategies reverse that order.
They start with capability and hope the workflow catches up.
It rarely does.
The real question leaders should be asking
Right now, the dominant question in many organizations is:
Which AI tool should we buy?
That is the wrong question.
The better question is:
Are our people ready for the work to change?
Ready does not mean everyone is technically fluent.
It means people can see their work clearly enough to separate routine tasks from judgment tasks. It means they understand where AI can accelerate the work and where it could damage quality, trust, or context.
That kind of readiness does not come from a demo.
It comes from preparation.
From sitting with the people who do the work.
From understanding the task flow, not just the org chart.
From showing people what changes, what does not, and what becomes more important once routine work speeds up.
This is where many leaders still get stuck. They talk about jobs at too high a level. They ask whether AI can do the role instead of asking which tasks inside the role are routine, which require judgment, and what happens when one part gets faster but the rest of the system stays the same.
That is the preparation work.
And it has to happen before the tool shows up, not after.
Why this matters now
This matters now because the technology side is accelerating fast.
The tools are getting better. Cheaper. Easier to access. Easier to test.
The human side is not moving at the same speed.
And every month leaders delay the preparation work, the gap gets wider.
The organizations that win in this next era will not be the ones that adopt AI first. They will be the ones that build the human capability to absorb change well.
That is a different kind of advantage.
Not speed for its own sake.
Readiness.
Clarity.
The ability to integrate a new capability without creating confusion, duplication, fear, or shallow compliance.
Preparing the humans is slower than installing software. It requires real conversation, real observation, and real diagnosis of how the work happens today, not how the SOP says it happens and not how leadership assumes it happens.
It also requires leaders to face something uncomfortable:
Most operations are already carrying more hidden fragility than they think.
AI does not remove that fragility.
It often exposes it.
What looks like an AI adoption problem is often an operating system problem that was already there.
AI just reveals it faster.
Do It Monday
Before your next AI meeting, do this with one role on your team.
Not a department. One role.
Ask three questions:
1. What does this person actually do every day?
Not the job description. The real work.
2. Which of those tasks are routine and repeatable?
That is where AI may help.
3. Which tasks require judgment, context, trust, or relationships?
That is where human value becomes more important, not less.
Write down your answers.
Then take them to the person doing the role and ask, “Did I get this right?”
That question alone will teach you more than most vendor demos.
If you can answer those three questions well, you have started to see the system.
If you cannot, you are probably not ready to buy another AI tool yet.
That is not failure.
That is clarity.
And clarity is where preparation begins.
The organizations that win in the AI era will not be the ones that moved fastest.
They will be the ones that prepared the humans first.
One thing I keep coming back to: most adoption problems show up as “people issues,” but they usually start as design issues. If the work, context, and expectations are unclear, even a good tool will struggle to stick. Curious what part of this feels most familiar in your world.