How should an organisation choose its first AI opportunity?
How to identify high-value AI opportunities before building anything
Start with a workflow where friction is real, measurable, and owned — not with a tool looking for a use case.
Name the operational friction
A useful AI opportunity starts with a specific process that is slow, repetitive, error-prone, costly, difficult to scale, or frustrating for a customer or team. Describe the current decision, hand-off, evidence, and consequence before discussing a model or platform.
Make value and feasibility visible together
A high-value idea must improve something that matters and be practical enough to test. Consider the likely operational impact alongside available information, process ownership, integration needs, acceptable risk, and the organisation’s ability to verify the result.
Choose a proof that changes the decision
The first proof of concept should answer a business question: does this workflow become faster, more reliable, more useful, or easier to govern? A demonstration without a clear adoption path rarely creates durable value.
Sources and further reading
AI transformation consulting — Kumash Shah Evidence-led case studies — Kumash Shah