Agentic AI for Knowledge Management in the Agriculture Community

Abstract

The integration of Artificial Intelligence (AI) into Knowledge Management Systems (KMS) offers new opportunities to address persistent challenges in agriculture, particularly in crop disease identification and treatment. A novel conceptual model of an Agentic AI-empowered KMS tailored for the agricultural community of farmers and farming experts is proposed. Unlike conventional decision-support systems, the model introduces agency through AI components capable of autonomously acquiring, retrieving, storing, and applying knowledge while engaging in continuous feedback with human experts. Multimodal interfaces, i.e., chatbots for natural language queries and image recognition for both text and visual diagnoses, enable farmers to contribute real-time field data. Knowledge graphs and Large Language Models (LLMs) mediate the transformation of inputs into validated, context-aware treatment and pesticide recommendations. The model is grounded in Activity Theory, providing a socio-technical lens to align user requirements, community participation, and role distribution with KM processes. The originality of this approach lies in combining agentic AI capabilities with collaborative knowledge exchange, creating a self-improving agricultural knowledge ecosystem. Anticipated contributions include enhanced decision-making, improved crop yields, and sustainable farming practices, establishing a foundation for future empirical validation and system implementation.