Keith Decker's TAEMS framework modeled the quantitative relationships among tasks, uncertainty, deadlines, cost and quality so coordination mechanisms could be designed around the environment agents actually faced.
Reid G. Smith's Contract Net Protocol made negotiation a concrete mechanism for allocating tasks among distributed problem solvers, creating a durable pattern for multi-agent coordination.
Edmund Durfee's work on partial global planning showed how distributed problem solvers could coordinate using incomplete, local and sometimes outdated views of one another's plans.
Les Gasser argued that distributed AI required genuinely social concepts of knowledge and action, while also helping build practical environments for experimenting with interacting agents.
How Victor Lesser’s work on distributed AI, coordination and the organization of complex AI systems helped establish the intellectual foundations of multi-agent systems.