Map What Your Company Can Actually Do With AI
Before another roadmap, spend an afternoon working out which parts of the business could realistically absorb AI at all.
Erin Moore
Fractional Chief AI Officer
An AI capability map is a one-afternoon exercise: list your core processes, score each on volume, rule stability, data availability and consequence of error, and you have a defensible picture of where AI could realistically help. It is deliberately cruder than a roadmap and far more useful, because it is built from what your company actually does rather than from what AI can theoretically do.
The exercise
List your processes — fifteen to twenty-five for most businesses. Then score each 1–5 on four axes:
Volume. How often does it run? Low volume rarely repays the effort however irritating it is.
Rule stability. Does it follow consistent rules, or does every case differ? Instability is the strongest predictor of disappointment.
Data availability. Is the information already captured in a retrievable form? If it lives in someone's head, that is a process change first — see data readiness for AI.
Consequence of error. What happens when the output is wrong? This does not disqualify anything; it determines how much oversight the use case needs.
Reading the map
High volume, stable rules, good data, low consequence — start here. This is the quadrant your first project should come from, regardless of where the biggest prize sits.
High volume, stable rules, good data, high consequence — real value, needs genuine oversight and a rollback path. Second, not first.
Unstable rules — leave alone until the process is agreed. Automating a disputed process multiplies the dispute.
Missing data — the project is capturing the data. Worth doing on its own merits, and it is not an AI project.
What the map is for
Two things. It stops the conversation being about tools, because it starts from your work. And it produces the shortlist that feeds use case prioritization, which is the sharper scoring exercise you run on the survivors.
It also surfaces something uncomfortable and useful: companies frequently discover that their most painful process scores badly on every axis, which explains why previous attempts failed and why the next one would too.
Keeping it honest
Do it with the people who perform the work, not just their managers. The documented process and the real one diverge, and the gap is where AI projects die. An afternoon with the actual practitioners produces a map you can trust; a workshop with leadership produces one that describes an idealised company.
Then re-score annually. Feasibility moves as tooling and data improve, and last year's "not ready" is often this year's obvious candidate. The AI maturity model covers the organisational side of the same question.
Frequently asked questions
What is an AI capability map? A scored list of your core processes rated on volume, rule stability, data availability and consequence of error — a picture of where AI could realistically help, built from your work rather than from vendor categories.
How long does it take? An afternoon, done properly with the people who actually perform the processes rather than only their managers.
Where should we start on the map? High volume, stable rules, available data, low consequence of error. Save the high-consequence quadrant for after your organisation has run one cycle.
How often should it be redone? Annually. Feasibility shifts as data and tooling improve, so items previously judged unready often become the obvious next candidates.
Further reading
Tagged with:
Ready to Automate Your Business?
Let's discuss how AI automation can deliver measurable ROI for your organization in 90 days or sooner.