Abstract

AI-enabled software-intensive systems increasingly operate as continuously evolving socio-technical artifacts driven by data, models, automation, and rapid delivery practices. Yet, many organizations still manage lifecycle governance, predictive quality assurance, automation ROI, and cybersecurity intelligence as separate workstreams, resulting in fragmented controls, weak traceability, and delayed detection of systemic risk. This paper proposes an end-to-end AI-based systems engineering paradigm that unifies (i) lifecycle governance through measurable objectives and evidence-based decision checkpoints, (ii) predictive quality assurance by integrating defect-risk prediction with automated validation evidence, (iii) automation economics via auditable time and cost savings models, and (iv) cybersecurity intelligence by mapping operational telemetry to threat-informed risk signals and outcome-based security governance. The paradigm is implemented conceptually as a continuous digital thread that ties requirements, architecture decisions, model behavior, test evidence, deployment gates, supply-chain integrity signals, and security posture into a closed-loop decision intelligence system. We present architecture, lifecycle processes, metrics, and an evaluation protocol suitable for regulated and high-impact contexts, including clinical practice deployments and critical infrastructure environments, emphasizing explainability, auditability, and operational usability.