Autonomous Agent

Agent calls external tools dynamically.

Number of steps considered in planning.

Decomposing goals into sub-tasks.

Interleaving reasoning and tool use.

Letting an LLM call external functions/APIs to fetch data, compute, or take actions, improving reliability.

A system that perceives state, selects actions, and pursues goals—often combining LLM reasoning with tools and memory.