Case & CodeA FIELD GUIDE TO AI
Start learning
Beginner

AI agent

Also: agents

In plain words

An AI system that takes steps to do a task. It may choose tools, search, or work with files. Its access should have clear limits.

An assistant working through a to-do list
  1. Open the folder of letters
  2. Read each letter and note its date
  3. Write the dates into a table
  4. Email the table to the client stop: needs your approval

A chatbot answers one message. An agent keeps going: it decides the next step, does it, looks at the result, and repeats. The important design question is which steps it may take alone and which must wait for you.

The fuller definition

An AI agent is a system that can select actions and use tools while pursuing a goal, within the capabilities and controls of its environment.

Go deeper

The agent loop
  1. ObserveRead the task, the current state, and the last result.
  2. DecidePropose the next action: a search, a file read, a calculation.
  3. ActThe harness runs the action, if permitted.
  4. CheckStop condition met? If not, loop again.

The model proposes; the surrounding software executes and records. Where the loop may act, and when it must stop and ask, are decisions made outside the model.

A concrete example

An inventory agent chooses which unread file to inspect next and records uncertain dates for review.

The distinction that matters

A named prompt is not automatically an independently running worker or a safe autonomous system.

Related terms

Where it is taught

Prompts, skills, workflows, and agents

Source and further reading

Anthropic · Building effective agents

The definition and example are written for this library. The link gives background, official documentation or legal authority as labelled. Legal scope depends on the jurisdiction and current text; product features depend on release and settings.

Lessons, terms, practical guides and sources ESC