International Journal of Science, Engineering and TechnologyAuthors: S Jeyalakshmi, A Akhash Kumar, SP Goutham, N Sethumadhavan, D Sheik Mohamed Rashid
Abstract: Agile software development methodologies face persistent challenges in maintaining consistency across user story creation, effort estimation, and test case generation. These inefficiencies often lead to project delays, quality issues, and resource misallocation. This paper presents AgileCopilot, an AI-powered assistant that leverages Large Lan- guage Models (LLMs), Retrieval-Augmented Generation (RAG), and role-based automation to streamline Agile workflows. The system inte- grates historical sprint data stored in MongoDB to provide context-aware assistance for Business Analysts, Developers, Testers, and Product Man- agers. AgileCopilot employs a novel role-specific prompting mechanism combined with similarity-based retrieval to generate accurate user sto- ries, estimate story points, create comprehensive test cases, and provide sprint insights. Our approach demonstrates alignment with UN Sustain- able Development Goals, particularly SDG 4 (Quality Education), SDG 9 (Industry Innovation), SDG 12 (Responsible Consumption), and SDG 17 (Partnerships), by promoting efficient resource utilization and knowl- edge sharing in software development teams. Early projections and trial outcomes indicate improvements in story consistency, estimation accu- racy, and test coverage compared to traditional manual approaches.
