Pervasive Integration of Autonomous Agents

Integration of applications is costly and cumbersome. We propose an approach to automate integration among pervasive, autonomous agents, in which internal process specifications are matched with in coming messages at runtime. The core of the approach is an efficient matching module, which we demonstrate to be working well for a simple game of Blackjack played between agents that have different implementations. Finally, we outline some strategies for dealing with errors which arise when the algorithm has to choose between alternatives.