What Is an Agent

Defines the perceive-decide-act loop and autonomy spectrum that separates agents from scripts and chatbots.

A first walkthrough of calling an LLM, giving it a tool, and watching it decide when to use it.

Contrasts four overlapping terms by autonomy, tool access, and how much a human stays in the loop.

Traces how the term shifted from marketing buzzword to a specific engineering pattern with real primitives.

A glossary covering tool calling, grounding, orchestration, context window, and agent loop terminology.

Ten habits for evaluating whether a problem actually needs an agent versus a simpler script.