Most assessment infrastructure was built on one assumption, that the unit of assessment is a discrete item. Item banks, authoring tools, scoring engines, and delivery systems all assume it. Skills-based assessment changes the unit from item to performance, and that shift ripples through the whole stack, increasingly demanding AI-native infrastructure with human-in-the-loop oversight built in. Pairing a credentialing perspective with an engineering perspective, this session starts where the strain shows first, the item bank, then maps what has to change across authoring, scoring, delivery, and review, grounded in a live simulation-based assessment program. Attendees leave with a framework for evaluating their own infrastructure against where their assessment model is heading, and a clearer view of the strategic and technical decisions they'll actually have to make.