Survey on Agile Practices and Technology Adoption

Backlog refinement is a critical process within Agile practices which often faces challenges like ambiguous user stories, prioritization difficulties, and cognitive overload among team members. Teams spend a lot of time in grooming user stories and refining them based on the client or business requirements and customer feedback. In this paper, we present an empirical study, exploring the integration of Generative AI (GenAI), specifically Large Language Models into backlog refinement workflows to address these challenges. We conducted a survey among Agile practitioners and identified the key pain points and challenges, explored benefits and reservations to adopting Generative AI with backlog refinement. We analyzed the survey data, which indicates that the practitioners are cautious but optimistic about using Generative AI in backlog refinement, especially for data-driven insights and reducing time spent in refinement sessions, with more skepticism around its role in enhancing team dynamics or prioritization and reservations concerning trust and data security. Based on these observations, we propose a framework for enabling GenAI in backlog refinements while preserving critical Agile principles and human oversight.