Table of Contents
Fetching ...

LLMs for Supply Chain Management

Haojie Wang, Jiuyun Jiang, L. Jeff Hong, Guangxin Jiang

TL;DR

This work develops the first domain-specialized SCM LLM built on a retrieval-augmented generation (RAG) framework to integrate external domain knowledge during inference. It constructs a supply-chain knowledge base, evaluates the model with SCM certification exams and beer-game tasks, and uses the LLM to simulate both horizontal and vertical supply-chain games, analyzing competition and cooperation under varied information structures and risk preferences. Key findings show that RAG significantly improves SCM task performance and that LLMs reproduce classical phenomena like the bullwhip effect while revealing novel behaviors, with information sharing mitigating the bullwhip for risk-averse and risk-neutral agents (with nuanced effects for risk-seeking). The results demonstrate the potential of AI-powered SCM research and decision support, and point to future work on scalability, richer domain knowledge bases, and more rigorous evaluation frameworks to bridge theory and practice.

Abstract

The development of large language models (LLMs) has provided new tools for research in supply chain management (SCM). In this paper, we introduce a retrieval-augmented generation (RAG) framework that dynamically integrates external knowledge into the inference process, and develop a domain-specialized SCM LLM, which demonstrates expert-level competence by passing standardized SCM examinations and beer game tests. We further employ the use of LLMs to conduct horizontal and vertical supply chain games, in order to analyze competition and cooperation within supply chains. Our experiments show that RAG significantly improves performance on SCM tasks. Moreover, game-theoretic analysis reveals that the LLM can reproduce insights from the classical SCM literature, while also uncovering novel behaviors and offering fresh perspectives on phenomena such as the bullwhip effect. This paper opens the door for exploring cooperation and competition for complex supply chain network through the lens of LLMs.

LLMs for Supply Chain Management

TL;DR

This work develops the first domain-specialized SCM LLM built on a retrieval-augmented generation (RAG) framework to integrate external domain knowledge during inference. It constructs a supply-chain knowledge base, evaluates the model with SCM certification exams and beer-game tasks, and uses the LLM to simulate both horizontal and vertical supply-chain games, analyzing competition and cooperation under varied information structures and risk preferences. Key findings show that RAG significantly improves SCM task performance and that LLMs reproduce classical phenomena like the bullwhip effect while revealing novel behaviors, with information sharing mitigating the bullwhip for risk-averse and risk-neutral agents (with nuanced effects for risk-seeking). The results demonstrate the potential of AI-powered SCM research and decision support, and point to future work on scalability, richer domain knowledge bases, and more rigorous evaluation frameworks to bridge theory and practice.

Abstract

The development of large language models (LLMs) has provided new tools for research in supply chain management (SCM). In this paper, we introduce a retrieval-augmented generation (RAG) framework that dynamically integrates external knowledge into the inference process, and develop a domain-specialized SCM LLM, which demonstrates expert-level competence by passing standardized SCM examinations and beer game tests. We further employ the use of LLMs to conduct horizontal and vertical supply chain games, in order to analyze competition and cooperation within supply chains. Our experiments show that RAG significantly improves performance on SCM tasks. Moreover, game-theoretic analysis reveals that the LLM can reproduce insights from the classical SCM literature, while also uncovering novel behaviors and offering fresh perspectives on phenomena such as the bullwhip effect. This paper opens the door for exploring cooperation and competition for complex supply chain network through the lens of LLMs.

Paper Structure

This paper contains 40 sections, 26 equations, 9 figures, 4 tables.

Figures (9)

  • Figure 1: Management professional ability performance.
  • Figure 2: Supplier game under the duopoly Cournot model.
  • Figure 3: Supplier game under Cournot model of multiple oligopolies ($n=3$).
  • Figure 4: Supplier game under Bertrand model.
  • Figure 5: Supplier game with asymmetric heterogeneity coefficient under Bertrand model.
  • ...and 4 more figures