AI Agents and Agentic AI in plant breeding: new frontiers, opportunities and challenges

Abstract

Plant breeding increasingly integrates diverse data sources, including genomic, phenomic, and environmental information, yet current artificial intelligence (AI) applications remain largely isolated analytical tools. Coordinating the entire breeding pipeline, from data collection and quality control to analysis, prediction, and decision-making, remains a challenge. Here we hypothesize that AI Agents and Agentic AI systems, which autonomously execute tasks using reasoning, memory, and tool integration, may orchestrate complex breeding objectives by coordinating specialized agents across multiple data domains. This approach has the potential to enhance breeding efficiency, reduce operational errors, and accelerate genetic gain by connecting previously fragmented processes. Although empirical validations remain forthcoming, the framework presented here provides a conceptual foundation for future implementation and evaluation. While technical, ethical, and institutional challenges exist, implementing such AI frameworks could support breeders in managing the complexity of modern genetic improvement programs, ultimately contributing to enhanced cultivar development and food security.