Abstract
Abstract Flood-induced disruptions pose significant challenges to supply chain recovery, particularly in fast-moving consumer goods (FMCG) sectors in developing economies. However, limited empirical evidence exists on how distinct artificial intelligence (AI) capabilities function as decision-support mechanisms for recovery resilience. This study examines how predictive modelling and forecasting, real-time response decision-making, and automated recovery optimisation are associated with recovery performance, where higher values indicate faster and more efficient recovery, with logistics process flexibility as a mediator. Using cross-sectional data from 268 managers and engineers in AI-enabled manufacturing firms in Nigeria, analysed via PLS-SEM, the findings show that predictive modelling and automated recovery optimisation are associated with logistics flexibility and recovery performance. In contrast, real-time response decision-making shows a weaker association and no significant link with logistics flexibility. The study positions AI as an embedded decision-support infrastructure and shows that supply chain resilience is more strongly associated with anticipatory and execution-oriented capabilities than with purely reactive responses, offering practical insights for designing effective AI-enabled recovery systems in disruption-prone environments.