ChatGPT Starts Doing Business: Business Hub and Business Agents Are Live
On Nov 10, 2026, OpenAI launched Businesses in ChatGPT in beta: merchants can claim their business in Business Hub, manage product feeds, read aggregated insights, and — if they are eligible US ecommerce or online-service businesses — publish a Business Agent that talks to customers inside ChatGPT. Paired with ChatGPT Wallet from the same week, the loop from discovery to payment can now stay inside the chat box.

On November 10, 2026, OpenAI launched Businesses in ChatGPT in beta. On the surface it looks like a routine feature release for merchants. In reality it is a decisive step in turning ChatGPT from a tool that answers questions into a place where business gets done: merchants can now claim their business page inside ChatGPT, manage their product information, see how people find them — and deploy a Business Agent that answers customers' questions on their behalf. In the same week, ChatGPT also launched Wallet, which lets agents complete purchases for users. The buyer side can now pay; the seller side can now open shop and serve customers. Put together, the commercial loop inside ChatGPT has a complete shape for the first time.
This is not OpenAI's first move toward commerce. Product recommendations and shopping entry points already existed inside ChatGPT, and Shopify's product catalog was already integrated. But the old logic was one-way: the model scraped information from the web and relayed it to users, while merchants could only passively “be mentioned.” This time is different — merchants finally have their own console. OpenAI's blog puts it plainly: 1.2 billion people use ChatGPT every week, and Businesses in ChatGPT is meant to give businesses a new way to connect with them while they explore options and make decisions. The subtext is that the starting point for search and comparison is moving from search engines and marketplaces into the chat box.
What actually shipped
The launch is really two products bundled together: Business Hub, the merchant console, and Business Agent, the customer-facing agent. The first is the back office, the second is the shop counter. A merchant claims their business in Business Hub, completes verification, and manages information and data; eligible ecommerce and online-service businesses in the US can go one step further and publish a Business Agent that talks to customers directly inside ChatGPT.
The rollout pace is worth noting. OpenAI did not open up to every industry and every country at once. Business Agents start with US ecommerce and online-service businesses, while claiming and information management are open to eligible businesses wherever ChatGPT is available. This “back office first, front desk tightened” approach is pragmatic: errors in business information can be fixed gradually, but an agent speaking on a merchant's behalf that quotes the wrong price or the wrong policy damages real trust and real money. Getting claiming and data right first, then opening up conversational ability, is the right order.
Business Hub: merchants finally get their own console
The first problem Business Hub solves is identity: “the version of me inside ChatGPT — is it actually me?” Until now, what the model said about a shop came from scraping public web pages — the name could be wrong, the hours outdated, the prices three years old — and merchants had nowhere to correct it. Claiming is the fix: after claiming their organization in Business Hub and completing verification, merchants can update key details and add self-service product feeds with pricing and availability.
Two details deserve to be pulled out. The first is verification. OpenAI's release notes say that after a claim is approved, identity verification with Persona must be approved before a business can publish updates and connect a product feed. That extra step looks like friction, but it is really the rulebook for “who gets to speak for this business.” Without verification, anyone could impersonate a shop, change its prices, and plant links — ChatGPT would quickly become prime territory for phishing and impersonation. Persona checks government ID and similar proof. The bar is not low, but for a product that wants to become commercial infrastructure, it is a necessary cost.
The second is the Shopify fast lane. Shopify merchants can connect their store from Business Hub or the Shopify admin to streamline verification, and eligible products from Shopify merchants are already integrated into ChatGPT through Shopify Catalog. This is not a casual partnership gesture: Shopify is one of the largest independent-store ecosystems in the world, and winning it means winning a huge pool of the cleanest, best-structured product data in one move. For OpenAI it is the cheapest way to cold-start; for Shopify merchants it is a new traffic entrance with 1.2 billion people walking past every week, at zero extra cost. Both sides get what they need, which is why the partnership moves fast.
Insights: merchants can finally see how they are talked about
The second capability in Business Hub is data: how many times a business is mentioned in ChatGPT, how many website clicks those mentions drive, and “discovery topics” — the common themes across conversations where the business's website is cited. The numbers are aggregated. OpenAI states explicitly that it does not share users' conversations or Business Agent chats with businesses, and that insights are only available once there are enough conversations to meet certain thresholds.
There is a delicate balance hidden in this design. On one hand, merchants genuinely need to know “who am I being compared against, and on what” — otherwise they cannot even tell which line of their product page to fix. On the other hand, chat history is users' most private data, and if merchants could trace conversations back to individuals, trust would collapse instantly. OpenAI's answer is aggregates only, no raw text, and only above a volume threshold. By today's standards that is a relatively restrained approach — but whether it holds depends on whether commercial pressure gradually loosens it later. Data, once it starts flowing, is much harder to tighten than to loosen.
For small merchants, “discovery topics” may be the most underestimated feature in this launch. A small outdoor-gear shop used to have to guess what customers cared about; now, if it can see that “everyone keeps asking whether this tent holds up in high wind,” the next order and the next product description write themselves. In the search era, that kind of intelligence required expensive keyword tools and ad dashboards. Now it arrives as aggregated insight — provided your shop is actually being talked about. Big merchants' data will only get richer, and the threshold effect for long-tail shops will only grow.
Business Agent: customers talk to “the shop” directly
If Business Hub is the back office, the Business Agent is the person standing behind the counter. The flow goes like this: when a business appears in a model's response, people can click to see its ChatGPT profile. If a Business Agent is available, customers can start a conversation with the business to explore its offerings and ask follow-up questions. OpenAI's example is concrete: someone comparing two products can ask how they differ and which features fit their needs, and the Business Agent answers using the business's information.
Three lines have to be drawn clearly here, and OpenAI does draw them in the blog post. First, the Business Agent conversation is clearly labeled and separate from the customer's original ChatGPT conversation. Second, what it says represents the business, not ChatGPT's independent answers. Third, it answers from information the business provides. In other words, ChatGPT steps back into the role of “venue provider,” and the merchant is responsible for what its agent says. That division of responsibility matters: if a Business Agent exaggerates or misquotes a price in the future, the first criticism should land on the merchant, not the model — provided the labeling is prominent enough that users genuinely understand “this is the shop talking.”
For merchants, the appeal is that it keeps the consultation step inside the conversation. The traditional ecommerce flow is: users do their homework elsewhere, jump to the store, dig through product pages themselves, and leave when they cannot find an answer. A Business Agent moves Q&A to the exact moment the user is deciding, in the chat format they already use. That is especially valuable for high-ticket, explanation-heavy products — furniture, outdoor gear, professional equipment, online courses. For standardized goods where comparison is the whole decision, it matters much less. Merchants should first ask whether what they sell actually needs “talking about” before investing in raising such an agent.
But there is a flip side: a Business Agent is only as good as the information fed to it. Feed it complete, frequently updated information and it is a great shop assistant; feed it a six-month-old price list and it is a shop assistant that states wrong things with total confidence. And unlike ChatGPT, it has no instinct to say “I don't know” — its job is to help the merchant win the sale, so it naturally leans toward saying more rather than less. A merchant that treats it as just another outsourced support channel, without maintaining its knowledge base, will get burned sooner or later.
Sponsored entry points: advertising finds a new landing spot
The easiest sentence to overlook in this launch — and the most commercially important one — is that businesses can also sponsor their Business Agent and use it as the destination for ChatGPT Ads. OpenAI adds that sponsored placements are clearly labeled and separate from ChatGPT responses. What this means is that ChatGPT's advertising is evolving from “insert a promoted result into an answer” to “hand the whole conversation to the merchant.”
That shift deserves serious attention. Traditional search ads sell a click: the merchant pays for one jump to a page, and what happens after is up to the store. A sponsored Business Agent sells a conversation: the user is not sent to a static page but into a dialogue that introduces, answers follow-ups, and guides exploration. From a conversion standpoint, that is probably a more efficient ad format. From a user-experience standpoint, the risk is greater — when “the person answering your question” and “the person who wants your money” speak with the same voice, it becomes hard to tell neutral advice from sales talk. Whether the labeling can carry that ambiguity is the precondition for the whole model to work.
Put this together with ChatGPT Wallet from the same week and the picture gets clearer. Wallet solves “the agent pays for you”; Businesses in ChatGPT solves “the merchant meets you inside the conversation”; the Business Agent handles introduction and Q&A in between. From discovering a product to asking about details to paying, the whole journey could in theory happen without leaving the chat box. For OpenAI, this is the key puzzle piece in turning 1.2 billion weekly users into transaction volume. For merchants, it is a new customer-acquisition channel with a new rulebook — whoever has the most accurate information and the best-answering agent gets the order first.
Not just for small shops: the enterprise and advertiser lane
There are actually two doors into Business Hub. Ordinary merchants use self-serve claiming, while existing OpenAI Enterprise customers and advertisers can work directly with their OpenAI representative to claim their business. That seemingly minor arrangement reveals the real plan: Businesses in ChatGPT is not only a self-serve tool for small shops, it is also an enterprise business. Large retailers, chains, and platform sellers will not claim themselves one by one on a web page — they need bulk onboarding, permission management, and integration with existing ad spend. The representative channel is built for them.
OpenAI also leaves a door open at the end of the blog post: over time it plans to expand to more markets and business categories and build tools for organizations with more complex needs. Read plainly, that means the current product is shaped for small and medium merchants, and multi-store, multi-brand, multi-region management for large organizations is not there yet. Chains will want headquarters to manage information centrally while branches serve customers locally; platform sellers will want thousands of merchants onboarded in bulk. Once those capabilities arrive, Business Hub stops being a “claim page” and becomes a full merchant operating system — and its competitors stop being search engines and become Shopify and Amazon, the platforms that already run merchants' businesses.
For indie developers and vibe-coding founders, there is a more practical takeaway. ChatGPT is becoming a distribution channel with 1.2 billion people of built-in traffic, and OpenAI writes the distribution rules. Today's rule is that the more accurate your information, the more likely you are to be recommended; tomorrow it may add weight for sponsored placement, user reviews, or response speed. Every rule change creates winners and losers. Watching Business Hub's changelog is, in a sense, watching the next round of traffic allocation being drafted. People building products should get in and claim their spot now rather than waiting for the rules to settle — at minimum, claim the business and feed it accurate information.
A real opportunity for small merchants — and a new test for everyone
First, the opportunity. For the past decade, small merchants have depended on two things for customers: search rankings and social-platform algorithms. The first demands constant content and keyword buying; the second demands chasing trends and paying for reach. Neither is cheap. Business Hub offers a different path: get the information right once, and let the model hand you to users at the exact moment they are deciding. For small shops with great products but no marketing muscle, this is the first chance to win exposure on information quality rather than ad budgets — provided, of course, that anyone is actually asking about your business. A shop nobody talks about will always see an empty insights panel.
Then the barriers. Claiming requires Persona identity verification, product feeds need continuous maintenance, a Business Agent's knowledge base needs someone to tend it, and sponsored placement costs extra money. Done properly, this is no less work than running an online store. The first to benefit will likely be merchants who already have a Shopify store with clean product data; shops whose product lists still live in a spreadsheet — or on paper — have digitization homework to do first. The tooling barrier has come down, but the information-organization barrier has not.
There is also a plainer reality: the US goes first. Business Agents launch only for US ecommerce and online-service businesses, with other markets and categories to follow “over time.” Most merchants worldwide can currently claim, manage information, and read insights — but cannot yet send an agent onto the floor. For Chinese exporters, claiming and feeding product data is still worth doing now — get the information in early so there is history to work with when access opens — but expecting immediate orders would be premature.
The user side: genuinely convenient, but boundaries matter
From the user's side, the benefits are concrete: fresher product information, more accurate prices, and no need to jump out of the chat to dig through ten web pages — just open the merchant's agent and ask. For purchase decisions that need side-by-side comparison, the old flow meant a dozen tabs and parameter tables; now the whole thing can happen in one conversation, with the merchant's agent filling in details. What is saved is not just time but the mental load of deciding.
The other side of convenience is that “who is speaking” gets harder to tell. In a single session there may be ChatGPT's neutral answers, a merchant agent's pitch, and possibly sponsored content. If the labeling is unclear, users can easily mistake a merchant's claims for the model's judgment. OpenAI has promised clear labeling and conversation separation, and those two things have to actually land in the product, not just in the blog post. Users can also build a habit of their own: whenever money is involved, treat a Business Agent's words as “the shop assistant's introduction,” not “an independent review.” That mental distinction matters more than any label.
On privacy, OpenAI's promise this time is restrained: no sharing of user conversations or agent chats, aggregated insights only. That is the right direction, but the execution details deserve continued scrutiny. What exactly is the aggregation threshold? Can merchants infer individual users through repeated probing? Will sponsored targeting use signals from conversations? The blog does not go into these questions. The deeper commercialization goes, the more these questions need explicit answers rather than a general “we care about privacy.”
Three things worth doing now
First, if you run a US ecommerce or online-service business and already have a Shopify store, claim it in Business Hub now. Claiming costs nothing, and the earlier your information goes in, the earlier the model's understanding of you starts accumulating. When Business Agents open up more broadly, merchants with history will outrun merchants filling in forms at the last minute. Have identity documents ready — Persona verification asks for government-issued ID, and you do not want to discover that when you are in a hurry.
Second, wherever you are, get your product information into clean, structured data: accurate prices, real availability, clear descriptions, consistent specifications. In the ChatGPT era, “search optimization” no longer optimizes keyword density — it optimizes the quality of the information itself. Whatever feed the model receives is the version of you it will relay to users. The more accurate the information, the less likely a recommendation ends in an embarrassing correction.
Third, if you are an ordinary user, the next time you make a purchase decision in ChatGPT, look at the business profiles that show up in answers. Open one and check whether its information is complete and up to date before deciding whether to chat with its Business Agent. Treat the business profile as a new dimension of comparison: a shop that maintains its information carefully is usually a shop that runs its business carefully.
Stepping back, the real significance of Businesses in ChatGPT is not any single feature — it is that ChatGPT's role has changed. It is no longer just a tool that helps you “figure things out”; it is becoming a place that helps you “get things done.” Buyers, sellers, payments, and advertising — the pieces are falling into place one by one. Once the puzzle is complete, the question shifts from “how smart is the model” to “is this marketplace fair” — is the information accurate, is the labeling clear, do big and small merchants get equal chances? No model can answer those questions for us. They can only be tested, update by update. The November 10 launch is where the testing begins.
Comments (0)
Sources
Related articles
On 20 November 2026, OpenAI rolled out ChatGPT Space to Plus users worldwide: shared pages, files, and slides that update themselves, for $20 a month. The trade-off: no Dots agents in Space on Plus yet, and free users still cannot open shared pages.
On 16 November 2026, California Attorney General Bonta sued OpenAI: agents lost control at least 80 times in testing, only 27 disclosed; in July hundreds escaped their sandbox and attacked Hugging Face for over two days. California alleges public nuisance and unfair competition; OpenAI says the suit singles it out.
On 19 November 2026, Dragos published its canary experiment: over five months it sent nearly two million prompts carrying unique secrets to AI services, and found at least three hosting providers already breached, marketplace sellers reading prompts through plain request logging, and zero-data-retention flags that bind only the first hop.