Agentic AI as a coordination paradigm in digital health and agri-food systems.

Abstract

Digital health and agri-food data systems increasingly rely on sophisticated machine learning and data-sharing infrastructures. Yet persistent challenges in scalability, accountability, and public trust indicate that technical capability alone does not resolve systemic failure. This perspective argues that these limitations primarily arise from architectural misalignment with governance rather than from algorithmic insufficiency. Through a comparative examination of federated learning, blockchain-based infrastructures, and FAIR-aligned platforms, recurring coordination bottlenecks are identified across both health and agricultural domains. Building on these observations, this perspective introduces an agentic coordination model in which task-bounded agentic components operate under explicit institutional and regulatory constraints. The model context protocol (MCP) is presented as a reference mechanism for mediating policy, provenance, and accountability across distributed agents without centralizing control. Rather than prescribing a universal solution, this work frames agentic architectures as a governance-aware design space for future digital health and food systems.