Weir, a lead proponent of artificial intelligence in mineral processing plants, highlights AI's potential to manage inherent variability in feed, ore grades, rock hardness, and mineralogy. Kenneth Ulrich, Head of Data and AI, explains that operators run equipment conservatively because variance is expensive, creating margins that AI can safely close. Weir’s NEXT Intelligent Solutions platform focuses on comminution improvements by modelling HPGRs, ball mills, hydrocyclones, and pumps. The approach uses academically-derived physical metallurgical models, constantly recalibrated to detect subtle process changes like shifts in ore hardness or moisture. AI models are trained with site-specific customer data due to process variability, requiring ongoing recalibration. For example, HPGR roller surface wear is difficult to measure directly, so a soft sensor—a software-generated signal mimicking a hardware sensor—is used to model wear. Benefits of NEXT Intelligent Solutions include predictive maintenance, process optimisation, and automated spare parts inventory management, tailored to customer requirements.