Maturing the field: from reasoning to action

1977 – Victor Lesser: multi-agent systems

In the late 1970s, distributed artificial intelligence (DAI) emerged. It was championed by Victor Lesser, who is widely recognized for pioneering multi-agent systems (MAS). His work focused on how independent software entities could cooperate, coordinate, and negotiate (see the Resources section). This development led to systems that were capable of solving complex problems collectively—an essential leap in building distributed intelligence.

1990s – Michael Wooldridge and Nicholas Jennings: the agent spectrum

By the 1990s, the distributed intelligence field had matured with contributions from researchers such as Michael Wooldridge and Nicholas Jennings. These scholars categorized agents along a spectrum, from reactive to deliberative, from non-cognitive systems to goal-driven, reasoning agents (Wooldridge and Jennings 1995). Their work emphasized that agents were no longer abstract ideas but were being applied across a wide range of practical domains, from robotics to enterprise software.

These researchers also introduced a shift in focus: from centralized reasoning to distributed action. Agents were no longer just thinkers—they were doers that operated in real-time environments with autonomy and purpose.

1996 – Hyacinth S. Nwana: formalizing the agent concept

In 1996, Hyacinth S. Nwana published the influential paper Software Agents: An Overview, which provided the most comprehensive classification of agents to date. His typology included attributes such as autonomy, social ability, reactivity, proactivity, learning, and mobility, and differentiated between software agents and traditional software constructs.

Nwana also offered a now widely accepted definition, paraphrased: A software agent is a software-based computer program that acts for a user or other program in a relationship of agency, which derives from the notion of delegation.

This formalization was instrumental in transitioning software agents from theoretical constructs to real-world applications. It gave rise to a generation of agent-based systems across fields such as telecommunications, workflow automation, and intelligent assistants.

Nwana's work sits at the convergence point of early distributed AI research and the operational architectures of modern agents. It is a crucial bridge between the cognitive theory of agents and their practical deployment in today's systems.