An AI agent is commonly understood as a system that pursues a goal across multiple steps, maintains state, and may choose or use tools to observe or change an environment.
Agent-like behavior has degrees. Planning, reading, writing, spending, sending, and continuing can each receive separate limits, approvals, monitoring, and recovery.
Example
Travel planning with controlled action
A system searches options, maintains a shortlist, and drafts an itinerary, but booking stays behind explicit approval and a spending limit.
Important limit
The label is unsettled and marketed broadly
A chatbot, fixed workflow, assistant, and agent can overlap. Inspect observable continuation, tools, state, permissions, action, and recovery.
Interactive model
Separate before deciding
Selected layerGoal pursuit
Continues across steps and updates state.
Read the complete text alternative
Goal pursuitContinues across steps and updates state.
AuthorityDefines what the system may read, change, send, or spend.
Useful distinctions
Separate the questions before judging
Goal pursuit
Continues across steps and updates state.
Authority
Defines what the system may read, change, send, or spend.
inAi