Software agents: Can we trust them?

In this paper we briefly describe multi-agent system architectures that are proved to be suited to address problems which simultaneously are of a Distributed, Decentralized and Dynamic nature. Two clearly different scenarios are used to show how the multi-agent system paradigm can be used to cope with realistic situations which are eiter mainly cooperative or mainly competitive and, thus, asking for specific characteristics that make those systems trustworthy. Negotiation protocols are used for the sake of getting, through cooperative strategies involving several agents, commonly accepted final solutions for an overall goal. We also advocate the use of Trust measures, together with normative environments that enforce legality, to make multi-agent systems useful in real demanding scenarios, like it is the case in competitive B2B situations.