Abstract
STI policies are key drivers of innovation but require context-specific approaches. While combining technology-push, market-pull, and systemic policies is widely advocated, empirical evidence on their effectiveness in agricultural innovation systems remains limited. This study applies the Adaptive Innovation System Model (AdaptISM), an Agent-Based Model grounded in prior theory, to analyse innovation in Antioquia, Colombia's coffee and avocado production chains - two sectors with distinct innovation patterns. Using empirical data for model validation, we test how different policy mixes affect innovation performance and economic outcomes. The analysis reveals trade-offs and synergies across policy instruments, offering practical insights for STI design. Our findings demonstrate how established innovation models can guide evidence-based policymaking in developing economies, particularly within diverse agri-food systems. This computational approach supports more targeted resource allocation and strategy formulation, grounded in local dynamics.