What Is an Intelligent Agent?

An intelligent agent is a computational system situated in an environment that is capable of flexible, autonomous action to achieve design objectives. The definition, formalized by Michael Wooldridge and Nicholas Jennings in their widely cited 1995 framework on intelligent agents, characterizes agents through four core properties: autonomy (operating without direct human intervention), reactivity (perceiving and responding to environmental changes), proactivity (exhibiting goal-directed behavior to achieve objectives), and social ability (interacting with other agents and humans). An intelligent agent perceives its environment through sensors and acts upon it through actuators, with the nature of those sensors and actuators depending on whether the agent is a software program, a physical robot, or an embedded control system.

The concept of the intelligent agent provides a unifying abstraction across artificial intelligence, control systems, and distributed computing. It situates reasoning and decision-making within an environmental context, emphasizing that an agent's behavior is always relative to the state of the world it perceives and the goals it pursues.

Autonomy and Agent Properties

Autonomy is the defining characteristic that distinguishes an agent from a passive software component. An autonomous agent controls its own decision-making process: it decides when and how to act based on its perception of the environment and its internal state, rather than executing commands issued directly by an external program. The degree of autonomy varies across agent types: a simple reflex agent selects actions based solely on the current percept using condition-action rules, with no memory of past states; a model-based agent maintains an internal representation of the environment that allows it to track aspects of the world not currently perceivable; a goal-based agent selects actions that advance toward an explicit goal state; and a utility-based agent selects actions that maximize a utility function representing the relative desirability of different outcomes. IBM's overview of AI agents surveys how these classical architectural distinctions map to practical implementations in modern AI systems.

Agent Architectures

The architecture of an intelligent agent specifies how its components: perception, reasoning, planning, and action relate to one another and interact with the environment. The Belief-Desire-Intention (BDI) architecture, originating in work by Michael Bratman on practical reasoning, models an agent through three mental attitudes: beliefs about the current state of the world, desires representing goals and preferences, and intentions representing committed plans currently being pursued. BDI agents have been deployed in applications requiring real-time decision-making under uncertainty, including air traffic control systems, automated trading platforms, and network management. Reactive architectures, by contrast, eschew explicit symbolic reasoning in favor of direct stimulus-response mappings and layered subsumption structures, as in Rodney Brooks's behavior-based robotics framework. Research on AI agent evolution and architectures surveys how classical agent architectures have been extended in the era of large language model-based agents.

Multi-Agent Systems

When multiple agents operate in a shared environment, the resulting multi-agent system exhibits emergent behaviors arising from agent interactions rather than from any single agent's design. Agents in such systems may cooperate to achieve shared goals, compete over limited resources, or negotiate to reach agreements. Multi-agent systems provide models for distributed problem solving, electronic markets, supply chain coordination, and collaborative robotics. The Jennings and Wooldridge survey of agent applications documented early deployments in telecommunications network management, air traffic flow management, and manufacturing scheduling, illustrating how agent-based decomposition scales to problems too large for centralized solutions.

Applications

An intelligent agent has applications in a wide range of disciplines, including:

• Autonomous vehicles and robotic systems navigation
• Automated trading and financial market modeling
• Network management and intrusion detection
• Smart grid and energy management systems
• Virtual assistants and conversational AI interfaces