The My Heart Counts study launches a comprehensive digital cardiovascular health investigation targeting 15,000 adults across the US and UK. Participants will contribute passive health data through smartphones—steps, heart rate, sleep patterns, ECG readings—while completing active fitness tests and clinical surveys. The embedded randomized trial compares AI-generated coaching prompts grounded in behavioral change theory against generic messages for boosting daily physical activity. The study represents an ambitious fusion of digital health monitoring and artificial intelligence intervention. This approach could democratize personalized cardiovascular prevention by eliminating geographic barriers and human expert bottlenecks that limit traditional coaching programs. The comprehensive data collection—from raw accelerometry to integrated electronic health records—may reveal novel digital biomarkers for cardiovascular risk prediction. However, the study faces inherent limitations of smartphone-based research, including potential selection bias toward tech-savvy participants and variable data quality across devices. The crossover trial design strengthens causal inference, though long-term adherence to AI coaching remains uncertain. As a preprint awaiting peer review, the study design may evolve before implementation. If successful, this model could transform how healthcare systems deliver scalable, personalized cardiovascular prevention at population scale.