Agentic artificial intelligence in food science: From automation to adaptation

This review synthesizes recent research on the emergence of agentic artificial intelligence in food, agricultural, and nutrition systems. While AI tools are increasingly deployed across domains from crop monitoring and processing to personalized nutrition and supply-chain optimization, current systems remain predominantly task-based, predictive, and narrow in scope. Recent scholarship highlights limitations in contextual reasoning, multi-objective decision-making, and the integration of ethical or governance constraints. This review organizes the literature across agricultural production, food processing, nutrition and health, food safety, and sustainability governance, and assesses how emerging agentic AI architectures may address existing shortcomings. Illustrative examples from food fermentation and personalized nutrition demonstrate persistent challenges when AI systems attempt to operate in real-world contexts characterized by social, ecological, and institutional complexity. The review concludes with implications for transparency, interoperability, and responsible governance, emphasizing the need for AI systems capable of deliberation, adaptive reasoning, and alignment with human and societal values.