Agent-based systems

Decentralized, autonomous software entities that perceive their environment, make goal-directed decisions, and execute multi-step actions without constant human oversight.

Agent-based systems (ABS) utilize multiple intelligent agents: each is an independent unit capable of self-sufficiency and real-time adaptation. These agents operate on a decentralized architecture, allowing for complex problem-solving that a monolithic system cannot handle. For example, in logistics or traffic management, agents coordinate dynamically to optimize flow, proving superior to traditional centralized control methods. Modern implementations often leverage LLM-powered agents for sophisticated reasoning, enabling them to break down complex goals and interact with external tools (APIs, databases) to complete end-to-end workflows in fields like finance and manufacturing automation.

https://www.ibm.com/topics/agentic-ai