🤖 AI Agent Classification: Architectures, Autonomy, Environments, Roles

Comprehensive overview of AI agents, defining them as computational entities that perceive their environment and act to achieve goals, exhibiting characteristics like autonomy, rationality, and goal-orientation. They establish a classical taxonomy categorizing agents by their internal architecture, from simple reflex agents to learning agents, highlighting the perception-action cycle and key architectural components. Furthermore, the sources introduce the modern paradigm of Agentic AI and Multi-Agent Systems (MAS), distinguishing a single AI agent as a building block from Agentic AI as a collaborative system of multiple agents. Classifications also extend to implementation and operational environments, differentiating between software and physical agents, API-based versus GUI-based agents, and their operational domains, alongside functional roles like information and transactional agents, and their autonomy levels which impact liability.