Large language models in supply chain management: a systematic literature review and application framework

Abstract

Modern supply chains face unprecedented complexity, volatility, and data heterogeneity, which challenge the effectiveness of traditional decision-making tools. Large language models (LLMs) offer a promising new approach to intelligent supply chain transformation with their contextual reasoning and semantic generalisation capabilities. However, existing research on LLMs in supply chain management (SCM) remains fragmented and exploratory, lacking a unified framework to guide theoretical development and practical deployment. Through a systematic literature review, the research identifies state-of-the-art applications of LLMs across SCM activities and proposes a structured framework for LLM-SCM applications. Guided by theoretical support and the Context–Mechanism–Outcome framework, the proposed framework maps LLM capabilities to the five core processes of the supply chain operation reference (SCOR) model, highlighting specific intervention points and application pathways. It demonstrates how LLMs can support resilient and insight-driven planning, ethical and sustainable sourcing, collaborative and traceable smart making, resilient and experience-driven delivering, and transparent and empathetic return. The framework not only enhances conceptual clarity on the role of LLMs in SCM but also provides methodological guidance for future research and practice. By aligning with the values of Industry 5.0, including resilience, human-centricity, and sustainability, the framework contributes to advancing intelligent and adaptive supply chain systems.