¿Qué es AI agent? Definición y Contexto

An AI agent is a language-model system that takes a goal as input and executes multi-step actions to accomplish it. Where a traditional chatbot replies once, an agent runs in a loop: read context, decide on a tool call, observe the result, decide on the next action, repeat until the goal is met or it gives up.

In coding, agents are typically given file system access (read/write), shell access (run commands), and a planning loop. The main constraints are tool quality (sharp, well-described tools beat sprawling APIs), context discipline (giving the agent only the files it needs), and stop conditions (knowing when the work is done or stuck).

Agents are most effective on tasks that have a clear definition of done — passing tests, matching a spec, fixing a reproducible bug. They struggle on ill-defined tasks ("make this nicer"), long-horizon planning, and anything that requires real product judgment.