Comment: JEL codes: D78, I32, D72, N16We use natural-language-processing algorithms on a novel dataset of over 900 presidential speeches from ten Latin American countries spanning two centuries to study the dynamics and determinants of presidential policy priorities. We show that most speech content can be characterized by a compact set of policy issues whose relative composition exhibited slow yet substantial shifts over 1819-2022. Presidential attention initially centered on military interventions and the development of state capacity. Attention gradually evolved towards building physical capital through investments in infrastructure and public services and finally turned towards building human capital through investments in education, health, and social safety nets. We characterize the way in which president-level characteristics, like age and gender, predict the main policy issues. Our findings offer novel insights into the dynamics of presidential attention and the factors that shape it, expanding our understanding of political agenda-setting.