The System Went Live. Nobody Changed the Work.

Early in my career, I was in organizational development at a large technology company, and I was part of the team rolling out a major Oracle implementation. Years of accumulated legacy systems, one integrated platform was meant to replace all of them. Serious budget, serious plan, and a steering committee that met more often than most of us would have liked.

A few weeks after go-live, I started walking the floors to see how people were settling in.

This was thirty-three years ago, and I cannot reconstruct a single conversation from it. What I remember is the pattern, because it repeated in every department I walked into. People would be perfectly polite, then slightly guarded, and then at some point somebody would open a desk drawer and pull out a yellow legal pad. Sometimes two or three people in the same room, once the first one had done it. Every number, every transaction they could not afford to lose, written down by hand.

Each department had quietly built its own backup system, and none of them had told us.

My first reaction, if I am being honest about it, was that these were people who did not want to change. It is an easy thing to think when you are one of the people who installed the system. It was also wrong, and it took me longer than it should have to see why. They were not being irrational. They were protecting their ability to do their jobs. The system was new and unproven, and its logic was invisible to them. The legal pad was familiar, it was visible, and it was theirs.

The platform was live, the project was closed, and the real work of the company was still running on paper in a drawer.

The drawer is the part that stayed, and it stayed because I keep meeting it again. Technology does not become transformation when it is installed. It becomes transformation when people trust it enough to change how they work.

The legal pad has changed shape since then. Now it is a private spreadsheet, a shadow approval process, a dashboard everyone opens and nobody acts on. The drawer is different. The behavior is exactly the same.

The question we stop asking on go-live day

Think about the last major rollout you were part of. On time? Probably. On budget? Probably.

Did the promised benefit actually land, consistently, in the daily work?

That one is harder, and most organizations stop asking it somewhere around the closing steering committee. Milestones have an owner and a date, which is what makes them easy to report. Whether people are now working differently enough for the business case to become real has neither, so it drifts out of the conversation exactly when it starts to matter.

And when the benefit does not materialize, it almost never arrives labeled as a people problem. It arrives as a forecast that is still off. A cycle time that did not improve. A maintenance alert that sat for two days. A procurement decision overridden by a manual check that nobody had asked for and nobody could quite explain.

Every implementation milestone was met. Not one habit moved.

Why AI raises the stakes

You could reasonably say the yellow legal pad is an ERP story, and that ERP is a solved problem. Fair enough. But I think AI makes the same problem harder, and the reason is worth naming, because it is not the one people usually reach for.

An enterprise system changes where the work happens. AI changes what the work is. It moves the decision itself, and in doing so it quietly renegotiates what counts as expertise, which tasks still carry status, and how much of the reasoning a person can actually see behind a recommendation they will be held accountable for.

So when someone hesitates over an AI output, they are rarely failing to understand the tool. No onboarding module is going to close that gap, because it was never a knowledge gap.

Motivation is not an ingredient you add

One framework I keep coming back to is Knoster’s model of complex change. Five conditions have to work together: vision, skills, motivation, resources, and an action plan. What makes it stick in your head is what it predicts when one of them is missing.

No vision, confusion. No skills, anxiety. No resources, frustration. No action plan, false starts.

And no motivation, resistance.

That last line is the one people quote, and it is also the one people misread. Read quickly, it makes motivation sound like a common ingredient, something a leader stirs in. Run a good kickoff, tell a compelling story, name the why, and the box is checked. But people do not move for the same reasons, and they certainly do not resist for the same reasons.

Picture a company introducing AI-supported planning. Two experienced planners, similar tenure, similar track record, sitting one desk apart.

The first one sees a way out of routine analysis and into the decisions she finds genuinely interesting. She is early to every session and asking for more access before anyone has offered it. The second built a twenty-year reputation on being the person who could read the forecast better than anybody else in the building. He is not confused by the system. He understands it perfectly, and that is precisely the problem.

From the outside, only one of them looks resistant. Underneath, they are not having the same experience at all, and the standard organizational response will reach exactly one of them.

More training will not solve a loss of meaning. More enthusiasm will not calm a fear of being exposed. More pressure will not create trust.

Both of them need to learn the same system. They do not need the same conversation.

The change is shared. Its meaning is personal. And that gap, between a shared change and a personal meaning, is where a great many transformation budgets quietly go to die.

Why we could never solve this before

None of this is new. It stayed unsolved because the solution did not scale.

The honest answer to “people need personal support through change” has always been coaching, and coaching is expensive, slow, and rationed. Which is why it tends to go to the top two percent, whose calendars already hold more support than anyone else’s in the building, while the change itself has to happen three levels down, in a thousand ordinary conversations that nobody senior is ever in the room for.

So we substituted. Group training instead of personal support. A communications plan instead of a conversation. A survey six weeks after the moment that mattered had already passed.

We were not being lazy. We were doing the only thing that was economically possible. That is the constraint that has changed.

What AI is actually good for here

Almost the entire conversation about AI at work is about getting people to adopt AI. Worth having. But it skips past something more interesting, which is that AI is the first thing we have that can make personalization economically boring.

Not personalization in the marketing sense, where your name goes in the subject line. Personalization in the sense of taking one organizational priority and translating it into what it means for this person, in this role, given what drives them, at the moment they actually have to act on it, without needing a coach present in every conversation. That is a translation problem at scale. It happens to be the thing this technology is good at, and the thing humans cannot do at volume no matter how much we care.

There is one condition attached, and it is not a small one. The AI has to be grounded in something real about the person, the role, and what the organization is trying to achieve. Otherwise, you have simply automated the all-hands message, and generic advice delivered faster is still generic advice.

This is the work we built Claro Mentor around. The mechanics are the useful part here, and the logic holds whether or not you ever use us. It works in four moves.

1. Map what actually drives each person. Not a personality type. Motivational drivers: what gives someone energy, what feels meaningful, what they act on first, what drains them. Thirty universal ones in our framework, each person’s top five forming their blueprint. Skill creates capability. Motivation decides how consistently that capability shows up in real work.

2. Ground it in the actual work. A driver map on its own is a nice conversation. Connected to this quarter’s priorities, this team, this role, it becomes guidance.

3. Deliver it privately, in the moment. Not a report that lands in an inbox. Practical coaching at the point of the hard conversation, the delegation, the decision someone has been avoiding for a week.

4. Give leaders patterns, not transcripts. Aggregated signals about where team energy is moving. This boundary is not a feature, it is the precondition. Personal coaching stays private, managers see patterns, and the moment people suspect otherwise, you are back to the yellow legal pad, except now they are hiding from you deliberately.

Back to the two planners. The first needs room to shape how the new process develops. The second needs a visible path to confidence and some evidence that his judgment still counts for something. Same rollout, same week, two completely different conversations that their manager now has a reason to know how to have.

The part that protects your credibility

None of this fixes a broken system, and pretending otherwise is how good tools get discredited.

If the process is wrong, if accountabilities were never defined, if the data is bad, no amount of motivational insight is going to help. It will just make people more articulate about why the thing is failing. Adoption work should never become a way of blaming people for a weak solution, and it gets used that way more often than anyone likes to admit.

A motivational layer earns its place when the business solution is sound and the gap is adoption, manager behavior, or how a team works together. When people know what to do and are not doing it. When training happened and nothing changed.

Being able to say “this is not the problem here” is what makes you worth listening to when you say it fits.

Where this leaves us

We have gotten good at the first mile. Selecting, implementing, integrating, going live. The last mile, the one that runs through what a person believes they are about to lose, is still mostly left to hope, a communications plan, and a manager who was handed a slide deck and no help. That mile is where the return on the whole investment gets decided.

So when someone tells me an AI rollout is complete, I no longer ask about usage rates. I ask what has changed that is not the system. Are decisions being made differently, or just faster in the same direction? Have the manual checks disappeared? That is the difference between installation and adoption, and it is the only version that eventually shows up in the numbers.

Technology can change the work. Only people can make the change work.

What is different now is that helping each of them find their own reason to move is finally something you can do at the size of your organization, rather than the size of your coaching budget.

Which leaves one question worth sitting with: somewhere in your organization right now, which rollout is still running on paper in a drawer?

Want to see what drives your people? Schedule a talk with Claro Mentor.

Get new articles by email: subscribe on Substack.