Self-Healing Retail Platforms Using Agentic AI: Autonomous Detection, Reasoning, and Remediation of Cloud Infrastructure Failures

Abstract

Failures in cloud infrastructures underlying retail platforms can lead to overly long downtimes, disrupting customer experience and risking loss of business to competitors. Self-healing cloud platforms are therefore needed that can autonomously detect failures, diagnose causes, and take corrective actions in production environments, returning to operation without human intervention. The idea stems from self-healing systems research, stemming from the artificial intelligence (AI) and agent-based systems (AS) fields. The fulfilment of key goals of self-healing systems—autonomous detection of failures, intelligent reasoning about causes and mitigation strategies, and execution of actions to restore operation—are outlined. Reflecting the Cloud Computing era, the focus is on cloud infrastructures supporting online retail platforms developed in the context of enterprise information systems curriculum. The self-healing capability is implemented using an agentic AI approach based on an intelligent agent framework that can integrate various AI technologies. Evaluation of the self-healing capability is based on practical diagnosis of cloud infrastructure failures by intelligent agents and quantitative assessment of detection and remediation speeds.

Failures in cloud infrastructures are detrimental to online retail systems. Recent work identifies Cloud Computing architectures and provides a catalogue of such failures. To mitigate long downtimes, triggered by the delays in human monitoring and diagnosis, the idea of self-healing systems is introduced. Recent research in self-healing systems shows that these systems can monitor and detect failures, reason about causes and mitigation strategies, and execute the suggestions. When the retail platforms supporting these business processes require dynamic adaptation due to changing user demands or the cloud infrastructure encounters failures, agent-based techniques allow the model to realise all of these dynamic adaptation requirements. The focus is on agent-based autonomous adaptation and self-healing capabilities using an agentic AI approach