An Action Research Study on Designing a Hybrid Project Management Process and AI Integration Strategy in a Software Startup

Abstract

Software development organisations have undergone rapid transformation over the past decade, particularly due to the rise of artificial intelligence (AI). However, project management practices have not evolved at the same pace, and companies often treat AI integration as a separate process rather than building on a solid project management foundation. This leads to inconsistent practices that limit the effectiveness of AI. This study addresses this gap by exploring how a hybrid project management process, combining PMBOK and Scrum, can be designed and enhanced through AI integration in a real-world software development organisation. The research adopts an action research approach in two phases. The first phase analyses existing project management practices using interviews, observations, focus groups, and ticket analysis, which reveals a lack of formalised process discipline. Based on this, process improvements are developed for the organisation. The second phase evaluates AI integration within the hybrid process. Testing of AI capabilities in ClickUp shows that performance depends heavily on data quality and consistency and reveals issues with the tool itself. Next, a root cause analysis is performed, which identifies organisational and process issues as key barriers to implementing AI in the organisation. Based on these insights, while measures are taken to improve the organisation’s project management processes, an in-house AI agent, Nydi, is developed to automate workflows such as daily updates and communication support in Google Chat. The findings show that AI agents improve efficiency but depend on well-defined processes and human oversight. The study validates PMBOK and Scrum as complementary frameworks, demonstrates a real-world application of agentic AI, and shows that effective AI integration requires an integrated strategy for both process and AI design.