Most AI governance is written to satisfy legal and read by nobody. The rules exist, adoption stalls anyway, and the organisation concludes people are the problem. They’re usually not — the rules just weren’t written for the person doing the work. Drawing on writing AI usage guidelines for a multi-market EMEA marketing organisation, this session makes the case for governance as an adoption lever rather than a brake: what it takes for guidance to meet the actual workflow, why the people expected to follow the rules should shape them, and how clinical protocols — followed voluntarily, by busy people, under pressure — offer a better model than the compliance memo.