INTEGRATION OF ARTIFICIAL INTELLIGENCE, AGILE METHODOLOGY, AND SOFTWARE ARCHITECTURE FOR IMPROVING SOFTWARE PROJECT SUCCESS

Abstract

The landscape of software engineering is evolving dramatically as a result of the merging of Artificial Intelligence (AI), Agile methodologies, and Software Architecture to meet the needs in today's world such as information overload and "guesstimation". This is especially significant in places such as Pakistan, where the IT industry is a worldwide freelancer leader and has a 40% failure rate of projects. The study highlights key findings, such as having Generative AI (GenAI) help with sprints can boost productivity by up to 88% and the AI-Based Architecture Decision Support System (AIDSS) gives architects 100% accuracy on the best tech options for the job. Moreover, the AIDSS framework is known to increase the efficiency of documentation as much as 40% of time and has been reported to achieve up to a 95% success rate for expert-validated decision making in prior literature, with this study's own experiment achieving 100% compliance on a smaller test set. This study aims to look into the possibility of introducing AI-based Decision Support Systems (DSS) and machine learning algorithms (Random Forest Classifiers and Linear Regression) to take project estimation from subjective "best guesses" to the successful "big data facts". The study suggests a strategy called "Agile Architecture" to reduce "architectural smells" and the “hit the wall” effect by having AI agents regularly check on architectural "smells". The study concludes that the "Human-AI Hybrid" approach should be adopted, with AI serving as a partner to maintain ethical standards and human creativity, while also complying with international norms such as IEEE 1471.