PostItFlow: An Early Study on Agent-Based Workflow for Enhancing and Visualizing User Stories




Weerakoon, Oshani; Rytilahti, Juuso; Mäkilä, Tuomas; Kaila, Erkki; Oyedeji, Shola

Scanniello, Giuseppe; Lenarduzzi, Valentina; Romano, Simone; Vegas, Sira; Francese, Rita

International Conference on Product-Focused Software Process Improvement

2025

 Lecture Notes in Computer Science

Product-Focused Software Process Improvement : 26th International Conference, PROFES 2025, Salerno, Italy, December 1–3, 2025, Proceedings

16361

561

570

978-3-032-12088-5

978-3-032-12089-2

0302-9743

1611-3349

DOIhttps://doi.org/10.1007/978-3-032-12089-2_42

https://doi.org/10.1007/978-3-032-12089-2_42

https://research.utu.fi/converis/portal/detail/Publication/508248731



Incomplete or missing requirements are a primary cause of project failure. During the initial phase of requirements elicitation, shortcomings frequently arise due to stakeholders’ differing perspectives and the predominance of elicitation methods that emphasize textual extraction. To address this challenge, we introduce PostItFlow, an AI agent-based workflow designed to provide a simple, easy-to-generate, and holistic visual overview of requirements. PostItFlow visualizes events and interactions among components, systems, and entities within a collection of user stories under an epic. Its objective is to reduce missed and incomplete requirements at the early stages of elicitation by offering a practical visual aid that can be utilized by both customers and requirements engineering (RE) practitioners. The visual output is produced through four progressive steps using GPT-5: (i) finalizing the given epic by adding supplementary stories where necessary, (ii) integrating a potential timeline that reflects the phases of the user story flow, (iii) enriching events related to entities or components with additional details, and (iv) generating the HTML-based visual output. We also demonstrate our workflow using a selected case study. This workflow will undergo industrial testing, with further enhancements guided by evaluations of usability, accuracy, and relevance to the information visualized. The prototype has been published as open source on Zenodo.



This work has been supported by FAST, the Finnish Software Engineering Doctoral Research Network, funded by the Ministry of Education and Culture, Finland.


Last updated on 28/01/2026 10:28:06 AM