Utilizing Large Language Models in Business Process Management: Applications and Challenges

Abstract

Large Language Models (LLMs) have recently been used in numerous domains, such as Business Process Management (BPM), which has significantly advanced. With LLMs' ability to understand language, reason, and tackle new challenges with minimal guidance, they offer an exciting opportunity to rethink and improve BPM practices. This systematic literature review examines insights from 42 peer-reviewed studies to understand how LLMs influence different stages of the BPM lifecycle. It sheds light on notable advancements and addresses the challenges that need to be overcome to unlock their full potential. Furthermore, we present an interactive Streamlit application that demonstrates the practical application of LLMs across all five stages of the BPM lifecycle using zero-shot learning, showcasing their potential to automate and enhance BPM tasks. We aim to deepen our understanding of the impact of LLMs on the evolution of BPM practices through a thorough review of current applications and future possibilities. The selected research papers cover LLM representation in various domains: process modeling (14%), process analysis and optimization (14%), process execution and monitoring (11%), process mining (19%), and generic capabilities and challenges (42%). Our findings underscore the growing importance of LLMs in addressing complex BPM scenarios while raising critical questions about scalability, interpretability, and fairness. Finally, this paper presents the technical, ethical, and practical challenges of integrating LLMs into BPM environments.