What is an AI roadmap?
By Joey van Vliet, founderUpdated
In short
An AI roadmap is a phased plan that sets out which AI applications an organisation tackles in which order, who owns each one, what needs to be in place for each phase and at which moments you decide to continue or stop. It is a decision document, not a wish list.
What goes in it
Per phase, and no more than that:
- Which application: one process, named specifically.
- Why this one first: the reasoning compared with the alternatives.
- Who owns it: a name, not a department.
- What needs to be in place first: data, access rights, a data processing agreement, a works council decision.
- How you measure success: one quantitative criterion, one qualitative and one for risk.
- When you decide: and that stopping at that point is a valid outcome.
Don’t look too far ahead
A three-year roadmap with twelve applications is usually wasted effort in this field. The tools change substantially within a year. More importantly, your first implementation changes your view of what makes sense next. What was step four on paper turns out to be unnecessary after the first pilot, or suddenly urgent.
So work with a short horizon and a fixed moment to adjust course. For an SME, nine to twelve months ahead is enough, with only the first quarter in detail.
Why it matters in practice
An AI project that delivers nothing often carries on because nobody dares to say it should stop. A roadmap that states upfront “after twelve weeks we decide on these three criteria, and stopping is a valid outcome” turns that decision into a normal moment instead of a loss of face.
That’s what sets a roadmap apart from a statement of ambition: it describes not only what you’ll do, but also what would make you stop. Why pilots without such a moment drift is covered in why AI pilots stall.
Frequently asked questions
How far ahead should an AI roadmap look?
For an SME, nine to twelve months is realistic, with only the first three months worked out in detail. Planning further ahead gives false certainty: the tools change quickly, and the first implementation almost always changes your view of what makes sense next.
What separates a useful roadmap from a nice slide?
Three things. It has names instead of departments. It has decision points where stopping is a real option. And for each phase, it states what must be in place before you move on. Without those three, it’s a wish list with a timeline around it.
