Agentic AI in Agriculture: Autonomous Systems for a Sustainable Future

Global agri-food systems are under growing pressure from climate volatility, resource constraints, and persistent labor shortages. At the same time, agriculture must increase productivity while reducing environmental impacts. We argue that agentic AI, defined here as goal-directed systems that integrate perception, reasoning, action, and learning in continuous closed loops, can help move digital agriculture beyond reactive optimization toward proactive and adaptive autonomy. In this Perspective, we present a unified framework for autonomy across agricultural scales, from in-field operations to supply chains. We also identify the key technical and system-level barriers that currently limit real-world deployment, and we outline a research path focused on scalable robustness, sustainabilitycentered objectives, and autonomy that remains under meaningful human governance.