Agile User Story Builder: Transform Real-Time User Requirement Conversations into Well-Structured Agile User Stories

Abstract

Agile software development methodologies rely heavily on well-structured user stories to bridge technical and business requirements. However, the traditional manual process of converting stakeholder conversations into structured user stories presents significant challenges including transcription errors, communication ambiguity, and inefficient processing. This research presents an automated Agile User Story Builder system that transforms requirement conversations into well-structured user stories. The system integrates cloud speech recognition, Natural Language Processing (NLP), machine learning, and Retrieval-Augmented Generation (RAG) for automated requirement processing within a microservices architecture. The system was developed using synthesized audio from 1,247 SRS documents (achieving 88% transcription accuracy) and validated on 75 real stakeholder conversations. Real-world validation demonstrates 78% speech-to-text accuracy (95% CI: 74-82%), 82% task classification precision, 73% reduction in manual documentation effort, and 86% format compliance in generated stories. The system achieves 20% timeline and 62% budget prediction improvement while providing measurable benefits including enhanced collaboration and streamlined Agile workflow efficiency.