Medical error is the third-leading cause of death in the United States, just behind heart disease and cancer. We describe a software platform used to train healthcare workers to prevent their errors. The platform (Amplifire) harnesses artificial intelligence and principles of cognitive psychology. Amplifire’s AI continuously decides whether and when to require additional learning events, provide corrective and metacognitive feedback, and/or deliver self-regulatory guidance for the learner (e.g., “slow down”). Amplifire was deployed to several thousand nurses at a large healthcare system in attempts to reduce the rate of two types of hospital-acquired infections. The result was a 48% reduction in central-line-associated bloodstream infections (CLABSI) and a 32% reduction in catheter-associated urinary-tract infections (CAUTI). These findings demonstrate the effectiveness of using cognitive science along with AI in an e-learning platform.