Ai agents in healthcare can improve care delivery, but they require disciplined program delivery. This paper outlines planning and governance controls that prevent safety ambiguity.