A shop can keep its window immaculate and still have no say in how a stranger describes what is inside. That has been the peculiar position of businesses appearing in chatbot recommendations. On 10 November, OpenAI introduced ChatGPT business profiles through a new Business Hub, allowing eligible companies to claim their identity, update details and provide product information. The question is whether a more accurate shop window inside an AI assistant will translate into better decisions for customers.
The new beta has two distinct parts. Businesses can manage the information that represents them in ChatGPT, while a narrower group can introduce conversational Business Agents that answer questions on their behalf. OpenAI's launch announcement says the latter option initially covers eligible ecommerce and online-service businesses in the United States. Claiming a profile is not the same as being granted an agent, and neither is a guarantee of appearing at the top of a recommendation.
How ChatGPT business profiles change the business listing
Think of a restaurant with an outdated menu, a retailer showing the wrong opening times or a service business whose prices have changed. An AI answer based on incomplete information can create a misleading impression before the customer ever visits the company's website. Business Hub offers a way to give OpenAI more current material at source, including organisation information and self-service feeds that can contain product prices and availability.
For small businesses, that could be an improvement over relying entirely on whatever information a model happens to encounter. It also creates fresh administrative work. A price change in the shop needs to be reflected in the feed, a discontinued service needs removing and an incorrect description still needs investigating. Verification helps establish who is authorised to speak for the company; it does not magically make the underlying data reliable forever.
OpenAI says an eligible organisation operating where ChatGPT is available may claim its profile. Existing enterprise customers and advertisers have a route through their account representatives, while supported Shopify merchants can connect their stores to simplify verification. The geographic and category restrictions for Business Agents are different. A British retailer should not assume that claiming a profile means a conversational sales representative will immediately become available to it.
A sales conversation with a name tag attached
The more consequential development is the Business Agent. When someone encounters a company in ChatGPT, the profile may offer a separate conversation through which the visitor can ask about products or services. Imagine comparing two bicycles and wanting to know which frame size a manufacturer recommends, or checking whether a particular subscription includes telephone support. The company could provide more specific material than a general answer.
There is a significant boundary here. The business agent speaks for the business, not as a neutral shopping adviser. It is supposed to be identified as such, and its conversation is separate from ChatGPT's independent answers. OpenAI's Business Agents policy requires businesses to represent themselves accurately, disclose material conditions, avoid invented prices or claims and refrain from pretending that the agent is an independent human adviser.
This distinction matters for consumers. A helpful sales assistant might be well informed and still have a commercial interest in the outcome. A chatbot that quietly blurred that line could make a company recommendation look more objective than it really is. The ability to ask follow-up questions is useful only if the customer knows whose information and incentives are shaping the reply.
What businesses can see and what they cannot
The dashboard also introduces aggregated information about customer discovery. OpenAI describes insights such as how often a business is mentioned, how many website clicks it receives and broad topics associated with conversations citing its website. That could give a company clues about why customers are finding it, or why they arrive with a particular misunderstanding.
OpenAI says it does not supply companies with users' original ChatGPT conversations or Business Agent chats, and that aggregate statistics are subject to thresholds. That is an important protection, although the announcement does not provide a public, universal method for independently auditing the figures. A firm should treat the figures as platform-provided indicators rather than a complete account of customer demand.
A second boundary concerns paid placement. Businesses may sponsor agents and use them as destinations for advertising, but OpenAI says those placements are clearly labelled and separate from the assistant's answers. It would be misleading to confuse paying for advertising with buying a favourable recommendation. Consumers need to be able to recognise both the independent answer and the sponsored route to a company representative.
The awkward problem of who gets discovered
A claimed profile may improve the information available about a business without solving the much larger question of visibility. A café can be accurately listed yet rarely appear in an answer. A well-known retailer may be selected because the system has more information about it. Small firms will want to know whether the new machinery makes discovery fairer, and what practical steps genuinely improve the quality of their representation.
Those questions overlap with previous LiveAIWire coverage of shoppers' trust in AI purchasing agents. Buyers are already being asked to judge systems that search, compare and sometimes act on their behalf. Adding a company-controlled conversation changes the experience again. It could correct a specification promptly, but it could also introduce a polished sales pitch at precisely the moment somebody wants impartial help.
There is also a difference between providing accurate facts and rewriting the substance of an independent assessment. An unhappy customer review does not become inaccurate simply because a merchant disputes it. Similarly, having a verified profile should not turn all information about a business into its own preferred account. The useful test is whether consumers receive better facts while retaining access to competing evidence.
A practical opening, rather than an instant gold rush
For a business owner, the sensible starting point is mundane: confirm identity, check contact details, make product feeds accurate and keep them updated. Examine the aggregated insights with care, especially if the figures are small. If an agent becomes available, test common customer questions and make sure answers mention limitations, eligibility and fees plainly. The uneven adoption of AI among smaller retailers is a reminder that not every business has staff available to maintain another digital channel.
Customers, meanwhile, should treat the profile as a direct source of company information rather than a replacement for comparison. Ask what the product costs, whether it is in stock and what is excluded. For a consequential purchase, verify the terms on the business's own website. A smooth conversation is not a substitute for a contract, nor evidence that the assistant has checked every competing offer.
OpenAI is moving from merely describing the commercial world towards hosting more of the conversation between shoppers and sellers. That could make everyday questions easier to answer. Whether it becomes a trustworthy marketplace of information will depend less on how friendly the agents sound than on how clearly the system separates facts, commercial speech and independent recommendations.
About the Author
Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, cybersecurity and the social impact of emerging technology. LiveAIWire is an independent, human-led technology publication using AI-assisted research, editorial production and original AI-assisted editorial illustrations under his direction.