Reflective Agents for Adaptive Workflows

Adaptation to changes in organizational procedures and business rules is a sine qua non for workflow management systems, if they have to be useful to organizations. This paper describes an approach based on workflow agents capable of managing dynamic changes in business policies. The two key aspects of our approach are (i) the agentification of the process engine, in the sense that this is viewed as controlled and executed by autonomous workflow agents capable of reacting and adapting to external changes, and (ii) the fact that these agents are reflective, in the sense that they can observe and thus modify their own behavior. The model is described in the context of an agent-based framework for coordination with reflective capabilities, and is evaluated against a well-known case of dynamic change.