Enhancing Procurement Processes in Supply Chain Management with Large Language Models

Supply Chain Management (SCM) faces challenges due to increasingly complex global operations and rapid decision-making demands. Procurement, a core SCM function, involves processing extensive supplier information under time constraints, limiting thorough bid analysis. Large Language Models (LLMs), such as GPT-4o (ChatGPT), offer advanced capabilities for language understanding and automation. This paper empirically investigates the effectiveness of LLMs in procurement, specifically in analyzing supplier offers. A controlled experiment compared GPT-4o with human professionals performing procurement tasks, including formal offer checks, information summarization, and supplier bid comparison. Results showed the LLM significantly reduced processing time while maintaining high accuracy, outperforming the average human participant but not the top-performing individuals. These findings indicate that LLMs effectively automate procurement tasks, enabling quicker and more informed decisions. Integrating LLMs into procurement can streamline operations, allow procurement experts to focus on strategic activities, and potentially enhance supplier selection outcomes. However, successful implementation requires careful consideration of biases, explainability, and operational risks.