AI-Driven Decision Intelligence for Agile Software Lifecycle Governance: An Architecture-Centered Framework Integrating Machine Learning Defect Prediction and Automated Testing

Abstract

Agile enterprise software teams are expected to deliver new functionality rapidly while maintaining high reliability and compliance. Traditional governance practices struggle to keep pace with the volume and complexity of decisions that must be taken across modern microservice-based systems. Prior work introduced a decision-intelligence methodology for AI-driven agile lifecycle governance and architecture-centered project management [1]. Another study discussed how machine-learning models are becoming pervasive artefacts across the software development lifecycle [15]. Reliability implications of automated testing frameworks in Java enterprise systems have also been examined [16]. Building on these foundations, this paper proposes a Decision-Intelligent Agile Governance Framework (DI-AGF) that integrates decision intelligence with machine-learning-based software defect prediction and automated testing. DI-AGF organizes the lifecycle into four layers: (i) observability and data engineering, (ii) analytics and machine learning, (iii) decision intelligence and policy, and (iv) execution and feedback. Governance decisions such as release readiness, test scope selection, and technical-debt remediation are treated as explicit decision assets that consume defect risk scores, architectural criticality metrics, and test evidence. The paper describes the architectural structure of DI-AGF, outlines how ML defect prediction and automated testing are orchestrated in a risk-driven manner, and illustrates the framework in the context of a Java microservices platform. An evaluation strategy and research agenda are also proposed. The contribution is a consolidated, architecture-centered reference model for operationalizing decision intelligence in agile software governance.