Cybersecurity threats targeting endpoint devices pose significant risks to the integrity and continuity of supply chains. As employees often represent the first line of defence, their ability to detect and respond to cyber incidents is crucial, but also highly variable, influenced by factors such as training, experience, and cognitive limitations. This study explores how artificial intelligence (AI) coaching can help in human decision-making in the context of endpoint cyber risk management. Comparative analysis of simulation results shows that an AI-coached model improves breach recognition, accelerates response behaviours, and reduces the likelihood of severe outcomes. The findings suggest that AI coaching is a promising strategy for strengthening human-centric cybersecurity practices in digitally integrated supply chains.