An AI use case is a precise description of a single task that artificial intelligence is meant to carry out, including what goes into it and what comes out of it. Most unsuccessful projects fail right here – the brief reads improve customer support instead of evaluate an incoming email, classify it into one of five categories, and draft a reply for the operator to approve. A usable use case contains a trigger, the input data and its source, the expected output and its format, who checks the output, what happens on error, and a measurable success criterion. Limit it to one process and one team. Rank use cases along two axes – benefit and deployment difficulty – and start with the one that is high on benefit and low on difficulty, even if it is not the most interesting one.
See also: Company AI Audit, Testing before investment, Return on AI investment.
