MBA Journals Peer-reviewed management research from Consortium e-Learning Network

NOLEGEIN Journal of Supply Chain and Logistics Management

A Cyber-Physical AI Architecture for Real-Time Inventory Control in Vendor Managed Inventory Systems

  1. Nitin Sharma
  2. Nikhil Kataria
  3. Anmol Bhatia

Abstract

Vendor-managed inventory is commonly employed to mitigate demand amplification in supply chains, but typically relies on static re-order policies and out-of-date information, making it relatively unresponsive to highly dynamic and uncertain conditions. While strides have been made in predictive inventory management using artificial intelligence and new sources of industrial sensing, most models are decoupled from the physical world and serve as decision-support tools rather than automated control systems. In this article, we propose a physical AI closed-loop inventory automation framework for VMI systems. The system uses a tightly coupled integration between real-time sensing, predictive demand modeling, adaptive replenishment logic, and actual fulfillment within a cyber-physical architecture to coordinate inventory decisions. The system leverages continuously updated feedback information from inventory consumption, lead-time variations, and replenishment outcomes to continuously learn — adjusting replenishment policies over time. This maintains human-in-the-loop transparency and supervisory control on an Industry 5.0 basis. The system then undergoes simulation studies to test it under a variety of demand scenarios and sudden supply interruptions. The new system is a major improvement over typical V.M.I. programs. “For those companies that adopt it, we’ve demonstrated that it reduces the number of stock-out events, reduces average inventory levels, increases the accuracy of inventory replenishment and overall resiliency of the system”, Bar-One said. This change from a reactive to a predictive mode of automation is enabled by predictive intelligence that is available through the physical internet. This research makes Physical AI achievable in inventory control and in turn unlocks a realistic path to human-centric, autonomous and resilient automation throughout the entire supply chain.

Keywords

Support