AI Agent

An AI that does not just answer, it acts: reads files, runs tools, makes changes, and works toward a goal across multiple steps with limited supervision.

All Dictionary

What it is

An AI agent is a model that can take actions, not just produce text. Instead of only answering a question, it can read your files, run commands, edit code, check the result, and keep going until a goal is met. Claude Code is an agent: you describe an outcome, and it works through the steps to get there.

Simple analogy

A chatbot is someone who answers your questions. An agent is someone who takes the task, does the work, and comes back with it done. The difference is acting in a loop, not just replying.

Why this matters for designers

This is the shift behind “agentic” design work. You move from asking for snippets to delegating outcomes: “build this section from the Figma frame and match our tokens.” Your job changes from writing every instruction to setting the goal, the constraints, and the quality bar, then reviewing the result.

How it works in practice

  1. You give the agent a goal and the context it needs.
  2. It plans, takes actions (read, write, run), and observes what happened.
  3. It iterates until the goal is met or it needs your input.