AI enabled real-time anomaly detection highlights unusual defect rates. As the system is self-learning, no process engineer need to manually tune thresholds.