OpenAI lets verified brands deploy chat agents inside ChatGPT

OpenAI is introducing a way for verified companies to operate their own representatives inside ChatGPT, according to reporting by The Verge. The feature, called Business Agents, lets users hold conversations with a brand's designated agent to handle tasks such as comparing products.

According to The Verge, the capability is launching in beta as part of a broader set of business-oriented features for ChatGPT. Access begins with e-commerce and online service businesses in the United States that OpenAI deems eligible.

A Business Agent becomes reachable when a company that has built one shows up in a ChatGPT response. From there, a user can ask about what the business offers and pose additional questions, including how two products differ. Companies control what the agent says by setting its instructions and supplying business context, per the report.

OpenAI says conversations with these agents will be clearly marked and kept distinct from the user's main ChatGPT thread. The point of the labeling is to make clear that the agent speaks for the company, and businesses can establish that status by claiming their organization and completing verification through a feature named Business Hub.

Specifics on how verification works remain undisclosed. Benji Shomair, OpenAI's VP of monetization, said during a press briefing that the company is not sharing details because verification methods differ depending on the business, The Verge reported.

Business Hub also gives companies visibility into how they appear in ChatGPT. Per the report, participants can view insights including mentions within ChatGPT responses and the leading discovery topics that surface their business.

In a blog post quoted by The Verge, OpenAI framed the launch as an initial phase. The company said it is starting with fundamentals such as accurate business information, useful insights, and conversations that support informed decisions. It added that it intends to explore further ways for businesses to reach customers, whether directly or through agents, both inside ChatGPT and elsewhere, with the aim of making ChatGPT more useful for people making choices and giving companies new avenues to participate and grow.

What this changes in practice is modest on the surface but potentially significant over time. For shoppers, it introduces a branded, conversational layer inside a tool many already use for research, though it remains to be seen how often agents appear in responses and how useful their answers prove. For businesses, it creates a claimed presence and an analytics view into how ChatGPT surfaces them, a form of visibility that has been difficult to measure.

Much remains unresolved. The report does not detail how eligibility is determined, what verification entails, or whether businesses pay for participation. It is also unclear how OpenAI will police agent behavior, how conflicts between an agent's claims and independent information are handled, or when the beta will widen beyond the initial US group. Because the source material is limited to OpenAI's announcements and a briefing, these gaps should be treated as open questions rather than settled outcomes.