Definitions
The definition of an AI agent, in one paragraph.
An AI agent is a software system that perceives information about its environment, decides on an action toward a goal, and carries out that action — then repeats the cycle using the result of its own actions, rather than waiting for a new prompt each time.
Breaking it down
The definition, one clause at a time.
"Perceives information about its environment"
It has access to current, relevant state — analytics, a codebase, a support inbox — not just the text of one prompt.
"Decides on an action toward a goal"
The goal is fixed ahead of time by a person; the agent decides which specific action moves toward it right now.
"Carries out that action"
It produces something real — a draft, a code change, a rendered video — not just a recommendation left for someone else to execute.
"Repeats the cycle using the result"
It checks what happened and folds that into its next decision, without a person re-prompting it.
What is NOT an AI agent
A single chatbot exchange doesn't meet the bar.
Typing a question into a chat window and getting an answer is one perceive-and-act step with no loop, no persistent goal, and no memory of the result. That's a useful tool — it's just not what "agent" means.
Questions
The things people ask first.
What is the definition of an AI agent?
An AI agent is a software system that perceives information about its environment, decides on an action toward a goal, carries out that action, and then repeats the cycle using the result of its own actions — rather than waiting for a new prompt each time.
How is an AI agent different from a large language model?
A large language model generates text in response to input; an agent is a system built around that model that gives it a goal, a memory of past steps, and the ability to take actions, then loops based on what happened.
Does an AI agent need to be fully autonomous to count as an agent?
No — an agent can still require a human to approve each action and count as an agent, because what defines it is the perceive-decide-act loop, not the absence of a human checkpoint.
What's the simplest example of the agent loop?
A marketing agent that reads this week's analytics (perceive), determines the underperforming post type is worth revisiting (decide), drafts a replacement (act), and checks next week's numbers before deciding what to do again (observe, repeat).
Get it before everyone else.
The download opens to the waitlist first, and the founding cohort is invited from this list. One email when it's ready — nothing in between.