Glossary

AI agent

An AI model put to work in a loop: it decides, acts with a tool, looks at the result, and repeats until the job is done.

What it is

A chatbot answers and stops. An agent keeps going. It takes an action, observes what came back, decides the next step from there, and cycles until the work is finished or it hits a stopping point. The simplest way to hold it: an AI agent is a model plus a harness. The model is the brain that predicts text and stops; the harness is the scaffolding that gives it tools, the loop, memory, and guardrails. That's why two products on the same model can feel completely different. The test for whether something is an agent: if a human hard-coded every step, it's automation; if the model decides the steps at runtime, it's an agent.

Why CEOs care

The chat window keeps you as the copy-paste layer between the model and the result. An agent removes you from that loop, so the work happens while you're in a meeting or asleep. The model is becoming a commodity; the harness around it is where the leverage lives, and the CEO running an agent against their own files pulls away from the one still pasting into a chat tab. Be honest about the limits, though: agents compound small errors across long tasks and run out of memory, which is why the reliable ones stay narrow and keep a human checkpoint before anything irreversible.

Where you'll see it

A coding agent (Claude Code) reading files, running tests, and fixing its own errors.
A support agent that reads a ticket, checks policy, and issues the refund.
A research agent that searches, reads, and writes you a sourced brief.

Full read

For the CEO-length version, see What is an AI agent?.

Workflows that use this

Go deeper

Put this term to work.

The Desk Theory guides turn definitions like this into running workflows. Two operator manuals, $99 each, or the bundle for $149.

Browse the books →

← All terms