A company buys a new AI tool. Leadership is excited. There's a kickoff, a login email and maybe a lunch-and-learn. Three months later, a handful of people use it every day and everyone else has quietly gone back to the old way.
I've seen this pattern with new software before, and I'm seeing it now with AI. The software is rarely the problem. The rollout plan covers the technology in detail and covers the people in one line.
Change has a sequence
I'm a Prosci-certified change practitioner, and the model I come back to most is ADKAR. It's simple on purpose. It breaks change into five elements, in order, and a person needs all five before a change sticks:
- Awareness that the change is happening and why
- Desire to take part in it
- Knowledge of how to do it
- Ability to actually do it in their real work
- Reinforcement so it holds once the launch energy fades
Most rollouts jump straight to knowledge. They schedule the training before anyone understands why the tool matters to them. Then they wonder why attendance drops after week one.
Start with the why, in their words
"We're adopting AI to drive efficiency" is a sentence for a board deck. It does nothing for the person who's worried the tool is there to replace them.
Before any training, I want every manager able to answer three questions for their own team. What is changing? Why now? What does it mean for me on Monday? If a manager can't answer those plainly, the team won't hear them at all.
Find the people who already use it
In every company there are a few people quietly using the new tool well. They figured it out on their own. They are your best training material.
Ask them to show their actual work in a team meeting. Ten minutes of a coworker saying "this is how I cut my Monday report from two hours to twenty minutes" does more than any vendor demo.
Plan for month three
The launch gets the budget and the attention. Month three is where adoption is won or lost. Put a check-in on the calendar before launch day: who's using it, who isn't, and what is getting in the way. Then fix what you find and tell people you fixed it.
None of this is complicated. It just has to be planned with the same care as the software.






























