This document discusses using a User-Trained Agent (UTA) to transfer knowledge between knowledge workers in an Adaptive Case Management (ACM) system. The UTA uses pattern recognition to observe knowledge workers' activities and learn from them. It stores what it learns in a central knowledge base and can then suggest the best next actions for knowledge workers based on similar past cases. Using business ontologies and negative learning examples helps the UTA learn more quickly and provide recommendations with higher confidence levels. The UTA aims to continuously acquire, share, and improve organizational knowledge without requiring specialized training.