Every AI roadmap I see starts at the wrong letter.
Companies come to me with the tool already picked, the budget approved, and a deadline. What they do not have is a sequence. So we jump straight to training: here is Copilot, here is the agent framework, here is the documentation.
That is Knowledge. And Knowledge is the third letter of ADKAR, not the first.
Before Knowledge comes Awareness (do people actually understand why this change is happening) and Desire (do they personally want it, not just tolerate it). Skip those two and the best training in the world lands on resistant soil. People sit through it, nod, and go back to their old way anyway.
Most AI rollouts do not fail because people are not smart enough or the tool is bad. They fail because nobody checked whether the team wanted the change before teaching them how to execute it.
ADKAR is sequential. You cannot buy your way past an earlier stage by spending more on a later one.
I wrote about this on my blog: https://culture-engineers.nl/blog/2026/10/29/your-ai-roadmap-is-missing-a-step-adkar/
A customer once asked me for an AI adoption roadmap. They already knew the tools, had the budget, and had a deadline. What they didn’t have was a sequence for the people who had to actually use it.
That request is becoming the norm. With AI, the tooling is so easy to buy that companies arrive with the solution already named, expecting the team to just pick it up once trained.
Sometimes that works. More often I see resistance, or something quieter and worse: people who nod in the workshop and go straight back to the old way afterward.
The model I reach for in these conversations is ADKAR: Awareness, Desire, Knowledge, Ability, Reinforcement. It is sequential, and most roadmaps start at Knowledge, the third stage, without ever building Awareness (the why) or Desire (do people actually want this, personally).
For every milestone on a transformation roadmap now, I ask one question: which ADKAR stage does this actually serve, and are the earlier stages done yet?
I wrote about this on my blog: https://culture-engineers.nl/blog/2026/10/29/your-ai-roadmap-is-missing-a-step-adkar/
When you built your last AI rollout plan, which came first: the training, or the reason people would want to change?
If it was the training, you are not alone. Almost every AI roadmap I review starts at Knowledge: here is the tool, here is the documentation, here is the office hours schedule. It skips straight past Awareness (do people understand why this is happening) and Desire (do they actually want to participate, not just comply).
ADKAR, the change model from Jeff Hiatt, lays out why that order matters: Awareness, Desire, Knowledge, Ability, Reinforcement. It is sequential. No amount of excellent training fixes a team that was never convinced the change was necessary.
People only start moving when they want to. So before you write the next training module, ask yourself: have you actually built Desire, or did you just assume it?
I wrote about this on my blog: https://culture-engineers.nl/blog/2026/10/29/your-ai-roadmap-is-missing-a-step-adkar/
Abstract portrait illustration of five stacked arches of a bridge, each arch rendered in a different material from rough scaffolding to polished stone, forming a path toward a light on the horizon, deep blue and warm amber palette, no text, 1024x1536 portrait format, geometric-conceptual style
Wednesday (per Mon/Wed/Fri rhythm), paired with the blog post publish date.