Glossary

AI assistant

A reactive, human-in-the-loop helper: you ask, it responds, and you steer the next move.

What it is

An AI assistant works turn by turn with you in the driver's seat. You prompt it, it answers, you read the answer and decide what to ask next. It drafts, suggests, summarizes, and then waits for you. The chat window, the in-app copilot, the writing helper baked into your email: all assistants. They're reactive by design, which is the point. You own the judgment at every step. Contrast that with an AI agent, which takes a goal and runs a multi-step loop on its own, deciding, acting, and repeating with little steering from you. The dividing line is where you sit: in the loop with the assistant, out of it with the agent.

Why CEOs care

The assistant is where you should start, and where most leverage still goes unclaimed. Before you hand work to an agent that runs alone, you want to be getting full value from the thing that keeps you in control. The assistant wins on judgment work: a board memo you own every word of, a pricing brainstorm where the value is the back-and-forth, a quick read of one document you can check. Anything where a wrong autonomous action would be expensive belongs here, not with an agent. Master the chat window first. Most CEOs get a fraction of what it can already do.

Where you'll see it

The ChatGPT or Claude window, where you ask and it responds.
An in-app copilot inside your email, docs, or CRM, suggesting as you work.
A Q&A helper answering questions over a single document you've uploaded.

Full read

For when each one fits and the trade-offs nobody mentions, see AI agents vs AI assistants.

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