Leveraging Generative AI in Agile Model-driven Development

Abstract

Despite growing interest in the effects and potential benefits of integrating AI techniques in software development, there is a significant gap in research that supports practitioners to leverage the technology effectively. This research identifies use cases for generative AI within Agile Model-Driven Development (AMDD) in such a way that it takes into account both strengths and limitations. To do this, multiple-case study research is performed, each case study focusing on a particular AMDD software project within a selected low-code development specialist & IT consultancy organization. Data for case analysis was gathered through semi-structured qualitative interviews with project members, along with the available project documentation and tooling. The identified use cases and their engineered prompts are validated through the expert opinion method by interviewing the project stakeholders and applying the use cases in their intended context. The results reveal a set of actionable use cases for generative AI, with prompt patterns and key insights for implementation. Additionally, this research presents three individual AMDD projects documented in process deliverable diagrams (PDDs), with an overarching, aggregated AMDD process deliverable diagram presented for broader insights. This research offers actionable insights and practical guidelines for practitioners who seek to leverage generative AI in AMDD, ultimately advancing the state of the art in software engineering methods.