Autonomous Procurement Systems for Resilient Global Supply Networks

Abstract

Global supply chains are increasingly exposed to disruptions arising from geopolitical conflicts, trade sanctions, climate-related events, logistics bottlenecks, supplier financial instability, and evolving ESG requirements. Traditional procurement systems remain largely reactive, relying on periodic supplier assessments, static risk scoring mechanisms, and fragmented decision-making processes that are insufficient for managing modern multi-tier supply networks.

This whitepaper introduces Autonomous Procurement Systems (APS), a conceptual research framework that integrates Multi-Source Risk Intelligence, Graph-Based Supply Network Analysis, Digital Twin Simulation, Multi-Objective Optimization, and Agentic AI into a unified architecture for resilient procurement decision-making. The framework proposes a five-layer model capable of continuously monitoring heterogeneous risk signals, modeling risk propagation across supplier networks, simulating disruption scenarios, optimizing sourcing strategies under uncertainty, and supporting governed autonomous procurement actions.

A key contribution of the framework is its explicit consideration of non-traditional procurement risks, including climate emergencies, warfare, sanctions, geopolitical instability, infrastructure disruptions, and systemic supply chain shocks. The paper introduces several novel concepts, including a Conflict Impact Propagation Model (CIPM), a Procurement Disruption Scenario Ontology (PDSO), a Graduated Autonomy Framework for Procurement AI, and a Synthetic-to-Real Data Strategy for future machine learning research in procurement.

Rather than advocating a specific algorithmic approach, the framework adopts an algorithm-agnostic research philosophy, identifying and comparing candidate techniques from machine learning, graph neural networks, operations research, reinforcement learning, digital twins, federated learning, and quantum-inspired optimization. The objective is to provide a rigorous foundation for future empirical research, enterprise innovation initiatives, and the development of next-generation procurement intelligence platforms.

This publication is intended as a foundational research artifact for academics, operations research practitioners, supply chain professionals, innovation teams, and enterprise architects exploring the future of resilient and intelligent procurement systems.

Keywords

Procurement AI; Supply Chain Resilience; Supplier Risk Management; Graph Neural Networks; Digital Twin; Operations Research; Reinforcement Learning; Agentic AI; Strategic Sourcing; Supply Chain Risk Intelligence; Geopolitical Risk; Climate Risk; ESG Procurement; Multi-Objective Optimization; Federated Learning; Autonomous Procurement Systems.

Authors

Somnath Banerjee, Subhamoy Bhaduri, and Binayak Mukherjee.

Version

Version 1.0 (Foundation Concept Paper), June 2026.