Agent vs Assistant
"Assistant" describes who a product is for — a person, being helped with something. "Agent" describes how it actually works — a loop that decides its own next action and keeps going without you. They're not opposites, and that's exactly why the marketing blurs them: a product can be called an assistant while being built as an agent underneath, or call itself an agent while barely deciding anything on its own.
What "assistant" describes
A framing around audience and purpose — software meant to help a specific person with a specific kind of task, the way a human assistant would. It says nothing about the mechanism underneath. An assistant can be a simple chatbot that answers questions, or a full AI agent that takes actions on the person's behalf — the word doesn't distinguish between them.
What "agent" describes
A specific mechanism: a system that decides its own sequence of actions toward a goal, rather than following a fixed script. It says nothing about who the product is for or how it's marketed — an agent can be built for consumers, for internal engineering teams, or for another piece of software entirely.
Side by side
| Assistant | Agent | |
|---|---|---|
| Describes | Who it's for, what role it plays | How it works, mechanically |
| Says anything about autonomy? | No | Yes — deciding its own actions is the whole definition |
| Can overlap with the other? | Yes — an assistant can be built as an agent | Yes — an agent can be marketed as an assistant |
| Best way to tell which it actually is | Ask who it's for | Ask whether it decides its own next step, or follows a script |
Why the confusion is so persistent
Because the two words answer different questions, a product can honestly be described by both at once — a "personal assistant" that's actually agentic under the hood, or an "agent" that's really just a scripted chatbot with one tool bolted on and a more exciting name. Neither word, on its own, tells you which one you're actually looking at. The real question — does it decide its own steps, or follow a fixed one — is covered on the Agent vs Chatbot page, which is about mechanism, not branding.
Which word actually matters for your decision
If you're evaluating a product to buy or build, "assistant" tells you almost nothing about what it can safely be trusted to do on its own — that's entirely a function of how agentic it actually is, not what it's called. Judge it the same way you'd judge anything calling itself an agent: how many steps happen without a person watching, and what it's allowed to do in those steps.
In this guide
FAQ
Is a "personal assistant" always less autonomous than something called an "agent"?
No. The name doesn't determine the mechanism — a personal assistant can be built with a full agentic loop deciding its own actions, and something branded as an agent can be a fixed workflow with almost no real decision-making in it. Check what it actually does, not what it's called.