Abstract: A system operative to execute and train at least one avatar for each user of an interactive environment comprising a knowledge engine operative to continuously monitor each user's response to events in the interactive environment while each user controls the at least one avatar, a knowledge base operative to store each of the monitored user responses to events in the interactive environment and an action engine operative to control one or more actions of the at least one avatar for each user in the interactive environment based on the stored monitored responses regardless of each user's control of the at least one avatar.