Agents as Webs of Beliefs

Instead of treating Beliefs, Goals, and Actions—the components that make up an intelligent agent—as separate elements, this paper proposes a new framework that models them all as Webs of Beliefs, all derived from the same underlying structure.

According to this approach, an agent's Reasoning, Planning, and Decision-making are not separate processes but naturally emerge from managing a network of beliefs that maintains local consistency.

This model moves away from existing fragmented approaches to agent design, providing a new theoretical foundation for building more flexible and organic intelligent agents.