As we observed in section 1, there has been a marked flowering of interest in agent technology since the mid-1980s. This interest is in part due to the upsurge of interest in Distributed AI. Although DAI encompasses most of the issues we have discussed in this paper, it should be stressed that the classical emphasis in DAI has been on macro phenomona (the social level), rather than the micro phenomena (the agent level) that we have been concerned with in this paper. DAI thus looks at such issues as how a group of agents can be made to cooperate in order to efficiently solve problems, and how the activities of such a group can be efficiently coordinated. DAI researchers have applied agent technology in a variety of areas. Example applications include power systems management [Varga et al., 1994][Wittig, 1992], air-traffic control [Steeb et al., 1988], particle accelerator control [Jennings et al., 1993], intelligent document retrieval [Mukhopadhyay et al., 1986], patient care [Huang and Fox, 1994], telecommunications network management [Weihmayer and Velthuijsen, 1994], spacecraft control [Schwuttke and Quan, 1993], computer integrated manufacturing [Parunak, 1995], concurrent engineering [Cutkosky et al., 1993], transportation management [Fisher et al., 1993], job shop scheduling [Morley and Schelberg, 1993], and steel coil processing control [Mori et al., 1988]. The classic reference to DAI is [Bond and Gasser, 1988], which includes both a comprehensive review article and a collection of significant papers from the field; a more recent review article is [Chaib-draa et al., 1992].