Agentic artificial intelligence of things (AIoT) for just-in-time (JIT) logistics and supply chain: a case of modular construction

Abstract

This paper presents a novel Agentic Artificial Intelligence of Things (Agentic AIoT) framework designed to overcome the critical challenge of achieving Just-in-Time logistics for bulky, costly modular construction. The framework merges autonomous, agentic artificial intelligence with Internet of Things infrastructure, creating a decentralised system for self-organising coordination among geographically dispersed stakeholders facing unpredictable disruptions. The research employs a mixed-methods approach, detailing a multi-layer architecture with specialised agents for data collection, intelligent search, multi-agent reinforcement learning, and autonomous control. A developed prototype was validated in two real-world modular construction projects, where it autonomously executed hundreds of dynamic schedule adjustments. The results demonstrate the framework’s transformative potential: it reduced inventory costs by 57.2% and 85.7% and significantly shortened disruption recovery times in the respective cases. This research establishes a foundational framework for developing autonomous, resilient logistics systems capable of intelligent coordination in complex, time-critical environments.