Internet of Agentic Things: AI Agents Orchestrate IoT Systems.

The Internet of Agentic Things (IoAT) proposes a novel architectural framework that unifies various advanced technologies to enable autonomous, closed-loop orchestration of physical environments. This framework combines agentic AI, the Internet of Things (IoT), cyber-physical systems, and edge computing, leveraging digital twins to create a comprehensive system. At its core, IoAT features autonomous AI agents operating across cloud, edge, and physical IoT layers. These agents are designed to perceive their environment, reason about situations, coordinate actions with other agents, and directly actuate changes within distributed cyber-physical systems. The paper formalizes this complex interaction as a coupled workflow-control problem, utilizing a hylomorphic dynamic programming approach to link agent planning with real-world execution. A key application highlighted is smart-building orchestration, demonstrating how IoAT can manage complex building systems autonomously. The research also delves into critical challenges for deployment, including ensuring safety, security, robust governance, system resilience, and overall trustworthiness of these highly autonomous systems.