Abstract
Realizing digital mental health’s promise in low-resourced settings requires a tool that works in real-world systems and that people engage with. We report a preregistered 6-week randomized controlled trial (N = 1,137) of Flourish, a scalable AI-powered mental health intervention, at a university in the Canadian province with the lowest psychologist supply. The intervention was introduced through academic infrastructure rather than offered as an opt-in wellness service. Participants were randomized to app access or care as usual. Flourish protected against declines in positive affect, resilience, and flourishing over the academic semester, and accelerated reductions in loneliness, extending prior evidence to a resource-constrained context and informing the emerging literature on whether purpose-built AI can improve well-being. Engagement was high: 95% activated the app within 2 weeks, and average use was 3.2 days per week. Embedding AI-powered well-being tools in institutional infrastructure may extend support where clinical capacity falls short of need.