Automating Supply Chain Disruption Monitoring via an Agentic AI Approach

TL;DR

This work introduces the first minimally supervised agentic AI framework for end-to-end disruption monitoring across extended multi-tier supply networks, integrating LLM-based reasoning with deterministic graph tools and a multi-agent workflow. Seven specialized agents orchestrate the process from unstructured disruption signals to multi-tier mapping, risk quantification, and actionable mitigation planning, all delivered as structured JSON outputs. Evaluation on 30 synthesized scenarios and a Russia–Ukraine case study shows high accuracy (F1 scores 0.962–0.991), rapid end-to-end analysis (mean 3.83 minutes) at minimal cost (mean $0.0836 per disruption), and substantial time-to-action improvements over manual, analyst-led approaches. The work highlights industrial readiness requirements and outlines future directions in real-time data ingestion, temporal networks, scalability, and human-in-the-loop governance to enable production deployment of proactive, autonomous disruption management.

Abstract

Modern supply chains are increasingly exposed to disruptions from geopolitical events, demand shocks, trade restrictions, to natural disasters. While many of these disruptions originate deep in the supply network, most companies still lack visibility beyond Tier-1 suppliers, leaving upstream vulnerabilities undetected until the impact cascades downstream. To overcome this blind-spot and move from reactive recovery to proactive resilience, we introduce a minimally supervised agentic AI framework that autonomously monitors, analyses, and responds to disruptions across extended supply networks. The architecture comprises seven specialised agents powered by large language models and deterministic tools that jointly detect disruption signals from unstructured news, map them to multi-tier supplier networks, evaluate exposure based on network structure, and recommend mitigations such as alternative sourcing options. \rev{We evaluate the framework across 30 synthesised scenarios covering three automotive manufacturers and five disruption classes. The system achieves high accuracy across core tasks, with F1 scores between 0.962 and 0.991, and performs full end-to-end analyses in a mean of 3.83 minutes at a cost of \$0.0836 per disruption. Relative to industry benchmarks of multi-day, analyst-driven assessments, this represents a reduction of more than three orders of magnitude in response time. A real-world case study of the 2022 Russia-Ukraine conflict further demonstrates operational applicability. This work establishes a foundational step toward building resilient, proactive, and autonomous supply chains capable of managing disruptions across deep-tier networks.

Figures (24)

• Figure 1: Overall stages, goals and data needed for autonomous disruption monitoring via an agentic approach
• Figure 2: Modular architecture of an autonomous agent. Each agent is composed of four key components: tools, memory, planning, and actions, which together enable it to reason, retrieve relevant information, and generate structured outputs.
• Figure 3: Proposed framework for autonomous agentic disruption monitoring. The framework consists of seven specialised agents (numbered blue boxes), each performing a distinct reasoning task in the end-to-end supply chain risk assessment pipeline. Outputs of each agent are shown in green and denoted as o-# for clarity and traceability across stages.
• Figure 4: Agent 1: Disruption Monitoring Agent - Input, output, tools and prompt
• Figure 5: Agent 2: Knowledge Graph Query Agent architecture, tools, inputs used and output structure