A discussion of two major benefits of using agents in software development

Abstract

Agent-oriented software engineering has not yet solved the basic problem of why we should use agents to build our software system. Why is it convenient to use agents instead of more mature technologies like, for example, software components? This paper addresses this issue and compares a BDI-like agent model with well-known component models like Enterprise JavaBeans, CORBABeans and .NET components. The two major results of such a comparison are: (i) agents are more reusable and more composable than components, and (ii) agents allow to describe systems at a higher level of abstractions than components. This work is not meant to be conclusive; rather it intends to start a debate on these and related topics.