AI Assistance in Agile Software Management: Real Impact or Hype

Abstract

Application of AI in agile software development has grown at a rapid pace, but there is little empirical evidence regarding the impact of AI on real Scrum teams in practical project environments. A lot of research that has been done is based on controlled experiments, tool-level testing, or theoretical discussions. Meanwhile, there are comparatively fewer studies that give a detailed, sprint-level analysis of how AI is applied in real Scrum team setups. This has led to a lack of knowledge regarding the impact of AI on productivity, workload, satisfaction, and team experience in actual Scrum processes. The thesis will fill this gap by exploring the effect of AI assistance in a project that has a Scrum-based approach.

This research has aimed at investigating the application and usage of AI assistance in Scrum with respect to productivity, perceived workload, team satisfaction, and general team experience. The main ideas that are considered are the AI-based productivity, perception of workload, team satisfaction, human control, and stakeholder value.

A case study research methodology has been used. The research has been anchored on a Scrum project, which has been executed within three sprints. Several sources have been utilized to collect data that will offer a holistic picture of AI usage. Some of these sources are project data that has been exported on the Scrum management tool, sprint sheets that are kept during the project, sprint surveys that are filled in by the team members after every sprint, and a further survey that was done with the owners and stakeholders of the websites. AI has been applied selectively in sprint activities where it was deemed useful by the team.

The results show that AI assistance has helped improve productivity through lowering the effort in certain tasks, especially those that are development-related and documentation-intensive. Perceived workload has increased over time despite continuous AI usage, suggesting that AI has shifted work from manual execution to review, coordination, and decision-making rather than eliminating workload. Team satisfaction has not been decreasing within any sprint, which is why the perceived usefulness of AI and team autonomy have had greater influence than the workload volume.

In conclusion, this thesis demonstrates that AI has a real but nuanced impact on Scrum. AI has functioned effectively as an assistive tool when integrated carefully into Scrum practices and governed by human judgment. This research gives useful information to Scrum teams and companies looking to adopt AI and adds empirical data about the impact of AI on actual Scrum teams