The new engineered systems are producing high volumes of sensor and operating data, which opens the possibilities of leaving behind the old models of reliability that are based on stagnant assumptions and scanty failure data. Simultaneously, industries are experiencing pressure in minimizing downtime, maximizing asset life and making cost efficient and proactive maintenance decisions. The proposed work advances a data-driven reliability modelling framework, based on artificial intelligence methods to produce failure predictions, remaining useful life estimates, and system performance and maintenance strategies. Supervised learning, deep learning-based prognostics and health management, and stochastic simulation are combined in the framework to learn the pattern of degradation using the multi-source time-series data. The reliability-centred metrics and optimisation processes are combined with feature extraction and training models to obtain risk-sensitive maintenance and operation policies. Industrial assets Case studies reveal that AI-enhanced reliability models have the capability to identify anomalies earlier, give more precise failure probability and remaining useful life estimations, and support predictive maintenance plans that considerably reduce unexpected unplanned downtime and maintenance expenses compared to the traditional preventive plans. The data-driven reliability modelling that can be provided by AI allows abandoning the reactive and time-based maintenance and adopting intelligent and condition-based decision-making, which can enhance system availability, safety, and lifecycle efficiency. The next direction in work will be model interpretability, uncertainty quantification, and incorporation of domain knowledge to make sure that it can be deployed in safety–critical uses with confidence.