Glossary

What is a branded agent?

A branded agent is a conversational interface, usually built on a foundation model, that represents one brand — letting users research, configure, shop, get support or transact by talking to the brand directly.

AI and conversational advertising

All 220 terms

What a branded agent is

It is the brand's own assistant: a chat interface on the brand's site, in its app, or exposed to other agents, that knows the brand's catalogue, policies, availability and pricing, and can act on them.

The important distinction is from a general assistant. A general assistant is neutral by design and compares you against competitors. A branded agent is openly partisan and everybody knows it — which is fine, because the user chose to talk to the brand. Nobody expects a shop assistant to recommend the shop next door.

What it is genuinely good at

Configuration and fit. Complex products where the real question is which variant, not which brand.
Post-decision support. Order status, returns, compatibility, warranty — high volume, low ambiguity, and expensive to staff.
Depth no general assistant has. Live stock, real pricing, account-specific terms.
Being callable by other agents. Exposed over MCP, it becomes the interface through which an agentic conversion completes.

What it does not do

A branded agent does not solve discovery, and this is the mistake most often made when one is funded. Users reach it after they have decided to consider the brand. The moment where a shortlist is formed happens somewhere else — in a general assistant, in search, in a conversation with someone else's agent — and no amount of investment in your own agent influences it.

That earlier moment is what AI share of voice measures, and it is moved by paid inclusion and by GEO / AEO, not by owning a chat window.

The liability that comes with it

Anything a branded agent says is the brand saying it. A general assistant that misstates your return policy is an inaccuracy about you; your own agent doing it is a statement by you, with the contractual and regulatory exposure that implies.

In practice this means constraining the agent to retrieved facts for anything commercially binding — price, availability, terms, eligibility — rather than letting it generate them, and logging what it said. This is the sharp end of hallucination liability.

Common questions

Will a branded agent get me recommended in ChatGPT?

Not directly. It can help indirectly by producing accurate, consistent, machine-readable information about the brand — but the recommendation decision is made by the general assistant's own sources.

Is a branded agent just a chatbot?

The difference is whether it can act. A scripted chatbot answers questions from a decision tree; a branded agent reasons over live data and can complete transactions.

Back to the glossary