Agents
AI agent
An AI agent selects and executes actions toward a goal using observed state.
How it works
The agent loop connects model decisions to tools and an environment. Reliability depends on state handling, permissions, stopping rules, and verification.
A concrete example
Illustrative: inspect 12 files, edit 2, run 3 checks, then stop when all checks pass.
Do not confuse it with Chatbot
a chatbot returns messages; an agent can take consequential actions through tools.