How Agentic AI Systems Work

Distinguish a foundation model, an LLM application, an assistant, and an autonomous agent.

Trace the perceive-reason-act-observe loop in a simple agent workflow.

Identify the roles of prompts, tools, memory, retrieval, and orchestration in an agentic system.

Classify levels of agent autonomy by the actions they may take without human approval.

Differentiate model errors, safety failures, privacy failures, and security attacks.

Diagram the components and data flows of a representative enterprise agent.