MULTI AGENT DECISON SUPPORT SYSTEM USING LLM's FOR CROP MANAGEMENT AND RESOURCE OPTIMIZATION

Modern agriculture increasingly depends on data-driven decision-making to improve crop productivity, reduce
resource waste, and ensure long-term environmental sustainability. However, traditional farm management
systems often operate in isolation, lack real-time intelligence, and fail to integrate the diverse data sources required
for precise decision support. To address these challenges, this project proposes a Multi Agent Decision Support
System (MADSS) powered by Large Language Models (LLMs) to optimize crop management and resource
allocation.
The proposed MADSS architecture is designed as a collection of autonomous yet coordinated agents, each
responsible for a specialized task such as soil monitoring, crop health assessment, weather forecasting, and
resource planning. By combining sensor data, farm records, and external knowledge sources, each agent uses
LLM-based reasoning to interpret data, generate insights, and provide context-aware recommendations. These
agents communicate through a central orchestrator, enabling collaborative problem-solving, adaptive learning,
and consistent decision generation across the system.A key strength of MADSS lies in its ability to process
dynamic, uncertain, and multimodal agricultural data. Through natural language interactions, farmers receive
clear suggestions on irrigation schedules, fertilizer usage, pest control measures, and yield predictions. The system
also supports real-time alerts, anomaly detection, and scenario simulation, helping farmers anticipate risks and
choose optimal strategies. Furthermore, the LLMdriven agents continuously improve through user feedback,
promoting long-term adaptability and scalability for diverse crop systems and field environments.
Overall, MADSS represents a modern, intelligent, and holistic approach to sustainable agriculture. By integrating
autonomous agents with advanced language models, the system enhances decision accuracy, minimizes resource
wastage, and empowers farmers with actionable insights. This project demonstrates how AI-based multi-agent
frameworks can significantly transform agricultural practices, moving toward more efficient, resilient, and
sustainable farm ecosystems