A shopper asking ChatGPT about a product while a business profile appears in the answer

For three years, the question businesses asked about ChatGPT was how to show up in its answers. On Tuesday, November 10, OpenAI gave them a door to walk through: Businesses in ChatGPT, a beta program that lets companies claim a verified profile inside the chatbot, correct the facts it shows about them, feed in products with prices and availability — and, for eligible American firms, put a branded AI agent of their own in front of shoppers, inside the answer itself.

It is the most direct move yet by a frontier lab to turn a chatbot into a commercial surface. OpenAI says ChatGPT now reaches 1.2 billion people a week. Until now, a business mentioned in an answer was a passenger in someone else’s sentence. Now it can own the listing, watch the traffic, and answer the follow-up.

What Business Hub actually does

Everything runs through a new portal called Business Hub, reachable from the ChatGPT sidebar. Any eligible business located where ChatGPT is available can claim its organization — at launch, only businesses with a website can claim — and, once verified, update its website, description, logo, products, prices, availability, and delivery and return policies. Retailers can add self-service product feeds so the facts ChatGPT shows stay current.

Verification runs through an identity check, and a claimed profile carries a blue check. OpenAI is blunt about one thing: claiming a profile does not guarantee higher placement in answers or more sales. The listing is a correction mechanism first and a marketing channel second.

The Hub also gives businesses something they have never had: a count of their own presence inside the model. The dashboard reports mentions — how often ChatGPT names the business — and website clicks over 7- and 28-day windows, plus “discovery topics,” the common themes of conversations in which the business’s site is cited. What it does not give them is the conversations themselves. OpenAI says businesses receive no user prompts and no transcripts, including chats with their own Business Agent. The counts are aggregated; the dialogue stays private.

The agent inside the answer

The sharper half of the launch is Business Agents. Eligible ecommerce and online-service businesses in the United States can build, test, and publish a conversational agent that chats with shoppers on their behalf. When the business appears in a ChatGPT answer, the user can open a separate, clearly labeled chat with the business’s own agent and ask it about products and services.

The limits are written into the policy, and they are deliberate. A Business Agent must be clear that it is AI, must not pose as a human or an independent adviser, and must not “create false urgency” or keep pushing a sale after someone says no. It is a pre-sale assistant, not a storefront: it cannot complete a purchase, make a booking, or check a customer’s order. The conversation hands off to the business’s own website for the transaction.

That restraint is the tell. OpenAI is not letting businesses close the sale inside ChatGPT — yet. What it is letting them do is own the last question before the click.

Ads are already built in

The launch also quietly wires the new surface into OpenAI’s advertising business. Businesses can sponsor their Business Agent and use it as the destination for ChatGPT Ads, OpenAI says — paying to put their agent where a shopper’s question lands. The hub and agents are free for now, and OpenAI has not ruled out charging for them later.

That makes the economics legible in a way the rest of OpenAI’s consumer business has not been. ChatGPT’s answers have been, until now, a space the company monetized only indirectly, through subscriptions. A claimable profile with a paid agent at the end of it is a marketplace model: the answer becomes inventory, the business becomes the advertiser, and OpenAI takes its cut at the point where a question turns into a purchase.

A crowded day, and a direction of travel

November 10 was one of the busiest launch days of the year. Anthropic opened its IPO roadshow the same day, pricing shares at $70 to $78 for a listing under the ticker ANT. Google made Gemini 4 Carbon generally available. DeepMind published AlphaProof Code, an agentic system that wrote more than six million lines of Lean to formally verify properties of Signal’s compiled code.

Read together, the day has a shape. The labs are no longer competing only on the model. They are competing on the surfaces around it: the workplace, the storefront, the verified proof. OpenAI’s move is the most commercially legible of the three. A chatbot that can answer “which one should I buy?” was always going to become a place where businesses pay to be heard. On November 10, it started charging rent — softly, in beta, with a blue check and a labeled agent, and with the sale itself still happening somewhere else.

What to watch next is how quickly the “somewhere else” disappears. The agent cannot take a payment today. The policy forbids it. But the infrastructure for it — the verified identity, the product feed, the sponsored placement — is now in place, and every piece of it points in the same direction.

Sources: OpenAI, “Introducing Businesses in ChatGPT” (November 10, 2026); Reuters, “OpenAI lets businesses claim ChatGPT profiles, adds AI agents for US e-commerce firms” (November 10, 2026); TechCrunch, Business Hub coverage (November 10, 2026).