A Supply Chain Network by Utilizing Multi-Intelligent Agents with Blockchain

Abstract

Modern supply chains face challenges including limited transparency, poor traceability, susceptibility to fraud, and slow responses to disruptions, particularly in traditional centralized systems. We present Bio-Cognito, a blockchain-integrated multi-agent framework that combines fuzzy neural networks (FNNs) with multi-agent systems (MAS) to enable predictive, adaptive, and secure decision-making across supply chain operations. Each agent coordinates tasks such as inventory management, supplier assessment, and logistics routing, while blockchain ensures data integrity and confidentiality. The framework integrates real-time multi-source data, including product inventories, supplier performance, and logistical operations, transforming it into actionable insights through FNN-based predictions and MAS coordination. Experimental evaluation on a Kaggle supply chain dataset comprising over 12,000 records shows that the logistics agent achieved 91% accuracy (Root Mean Squared Error (RMSE) 0.2), the supply agent 83% accuracy (RMSE 0.16), and the inventory agent 75% accuracy (RMSE 0.14). Ablation studies highlight the importance of multi-agent coordination, knowledge integration, and multimodal data fusion. Empirically derived thresholds enable phase-specific interventions, supporting reliable and interpretable decision-making. The proposed framework achieves high cross-domain generalization while maintaining real-time inference on standard clinical GPUs. Overall, it offers a systematic methodology to integrate intelligent agents with blockchain, enhancing transparency, predictive reliability, and operational efficiency in dynamic supply chain environments.