ChatGPT Business Hub and the Future of Commerce: How Brands Win in 2027 and Through 2030


Rahul Vijaywargiya
Nov 12, 2026 · @Rahul
In brief
What it is. ChatGPT Business Hub launched in beta on November 10, 2026. It lets a business claim its profile in ChatGPT, upload a product feed, and see its mentions and clicks.
Why it matters. It is the clearest sign yet that AI assistants are becoming a commerce channel and not only a research tool.
What it lacks. It covers only ChatGPT, shows no questions or competitors, and does not improve a brand’s ranking.
Who saw it coming. Bodhium Labs co-founder Kartik Hosanagar wrote in July 2025 that shoppers would ask LLMs before buying and that brands would need AI agents of their own. Both are now real.
What comes next. In 2027 the buying decision moves into the assistant. By 2030, AI agents complete a large share of purchases.
How brands win. A clear position, facts a machine can read, proof from independent sources, a strong brand agent, and testing content before it is published.
Where Bodhium Labs comes in. Bodhium Labs helps brands set up and run their presence on ChatGPT Business Hub, and improve their visibility and sales across every AI platform.
A new front door for commerce just opened
On November 10, 2026, OpenAI opened ChatGPT to businesses. Any eligible company can now claim its place inside ChatGPT, correct what the assistant says about it, load its products with live prices and stock, and see how often it gets mentioned and clicked. OpenAI says 1.2 billion people use ChatGPT every week.
It would be easy to file this as a product update. We think it is the most important opening a brand has had on an AI platform so far. For two years, brands could only watch what ChatGPT said about them. Now the largest assistant has given them a way in, and the other assistants are building the same.
The AI assistant is turning into the place where buying decisions get made, and it will soon be where many purchases get finished. Search sent shoppers to a shelf. The assistant picks from the shelf for them.
Bodhium Labs helps brands find their way through this new channel. That means getting set up on Business Hub the right way, and then improving how often and how well AI assistants recommend them. This article explains what changed, what happens next, and what a brand should do about it.
Bodhium Labs saw this coming in 2025
On July 30, 2025, fifteen months before Business Hub, Bodhium Labs co-founder and Wharton professor Kartik Hosanagar published “Hey AI, buy my product”. He described three stages: search, then answers, then agents. He wrote that shoppers would consult LLMs before buying, that answer engine optimization would become the main marketing battleground, and that businesses would need AI agents of their own to speak for them. Plenty of people expected AI to change shopping. Few said in which order it would happen.
What Kartik Hosanagar wrote | When | What has happened since |
|---|---|---|
Shoppers will ask LLMs before they buy | July 2025 | In a 2026 McKinsey survey of 21,000 consumers in 11 markets, 53% had used AI for shopping in the past three months |
Brands must track how they show up in AI answers | July 2025 | In November 2026, OpenAI began reporting mentions and clicks to brands |
Businesses will need their own AI agents | July 2025 | In November 2026, OpenAI launched Business Agents |
Retailers risk becoming fulfilment suppliers to AI platforms that own the customer | The fight over who controls the checkout defined 2026 |
In January 2026 he put it in one line for Retail Dive: “Whoever controls the agents now has the power.” The outlook in this article comes from the same research.
What OpenAI launched, and what it leaves out
Business Hub gives a brand four things inside ChatGPT: a verified profile, a product feed, a first set of numbers and, for some, an agent of its own.
What you get | What it does |
|---|---|
Verified profile | You claim your listing, pass an identity check, then edit your description, category, logo, delivery and return policy. A blue check shows the profile is claimed. (details) |
Product feed | You connect a Shopify store or upload a file. Each product carries its title, price, availability, image and link. (feed spec) |
Insights | You see how often ChatGPT mentioned you, how many clicks went to your site, and the topics you were cited for. (TechCrunch) |
Business Agent | US ecommerce and online service businesses can launch a labelled chat that answers product and policy questions. It is free for now and cannot take a payment yet. (Digiday) |
What it leaves out is where brands will get caught:
It covers one platform. There is nothing on Gemini, Google AI Mode, Perplexity, Claude, Copilot or Meta AI.
It shows no questions and no competitors. You learn that you were mentioned. You do not learn which questions you won, which you lost, or who was recommended ahead of you. Share of voice is not in the beta.
It counts only part of your presence. For online businesses, a mention counts when ChatGPT cites your website. Recommendations that draw on reviews, retailers or press about you are not in the number.
It does not move you up. OpenAI’s position is that claiming a profile does not guarantee better placement, and that ad spend does not change how often ChatGPT recommends a business.
Business Hub tells you the score. It does not tell you why, and it does not tell you how to change it. The recommendation is still earned outside the hub: in your content, your product data and what the rest of the web says about you.
The year of reluctance: why ChatGPT commerce stalled first
OpenAI’s first push into commerce did not work. What has changed since then explains why this one will.
In September 2025, OpenAI launched Instant Checkout, which let people buy inside the chat. The promise was more than a million Shopify merchants. By February 2026, about 30 were live. Walmart, the flagship retail partner, said purchases made inside ChatGPT converted at a third of the rate of clicks through to its own site. On March 24, 2026, OpenAI pulled the feature back and moved checkout to the merchants.
Three kinds of reluctance sat behind those numbers:
Retailers would not hand over the customer. Checkout inside someone else’s app meant giving up loyalty programs, promotions, baskets and customer data.
The biggest marketplace stayed out. Amazon blocked OpenAI’s shopping crawlers in late 2025 and backed its own assistant, Rufus.
Shoppers were not ready to pay in a chat window. They used ChatGPT to research and compare, then went elsewhere to buy.
Many people took this as proof that AI commerce was hype. We read it differently. OpenAI was too fast ahead. It asked for the checkout before it had addressed the concerns any brand would have: who owns the customer, who holds the data, and what happens to loyalty and promotions. Shoppers, meanwhile, never stopped asking ChatGPT what to buy.
What has changed since then is the brands. They now see this as an inevitable channel. The proof is in what they did after the pullback. Target, Sephora, Best Buy, Home Depot and others put their catalogues into ChatGPT’s product discovery, and Walmart placed its own assistant inside ChatGPT and Gemini. A brand that stays away will suffer lost customers, lost sales and lost reputation, because the conversation about its products carries on without it. Winning this channel has become very important, and with that comes a new openness. Brands are starting to accept some loss of control as an inevitable state.
Nobody knows yet what the final state will look like. We suspect that, eventually, a lot of the control will sit at the final surface, which is the LLM: the customer’s preferences, the conversation and, in time, the payment. The product, the price, the delivery and the service stay with the brand. Business Hub is an early step in that direction and a cheap one to take, since a brand gives up nothing it has today by claiming a profile. That is why we expect adoption this time to look nothing like the last.
From discovery to decision: how the funnel moved onto AI platforms
Since 2024, AI platforms have taken on the buying journey one step at a time, with one false start along the way.

buying journey on AI platforms · 5 steps, 2024 to 2030
To see what the chart means in practice, take a parent shopping for a stroller that is light, fits in an overhead bin and costs less than $400.
In 2024, that parent asked an assistant, got a few general suggestions, and went to Google and Amazon to do the real work.
In 2026, the answer names three strollers with current prices and stock, pulled from product feeds. The parent taps one brand’s profile and asks its agent whether the stroller fits the bin on a budget airline. Then they click through to the brand’s site to pay. Discovery and comparison now happen in the assistant. The decision is halfway in. The purchase is not, because OpenAI’s one attempt to move it there failed.
In 2027, the last doubts get settled in the same conversation, and the parent can pay there too. Once the questions are answered in the chat, the decision is made in the chat.
By 2030, the parent may not ask at all. Their assistant knows a flight is booked for March, knows the budget, and orders the stroller.
The top of the funnel took three years to move. We expect the bottom to move faster, because the habits, the feeds and the payment rails are already in place. For a brand, that means the page where it used to win the sale is no longer where the sale gets decided.
The signals behind the shift
Business Hub did not arrive alone. Three kinds of evidence point the same way.
First, shoppers are buying, and their visits are worth more. Among US consumers who use AI, the share completing a purchase with it rose from 10% to 17% in six months, according to McKinsey. Adobe found that by May 2026, visits arriving from AI assistants converted 54% better than other visits. A year earlier they converted at about half the rate.
Second, every platform is building the same thing. Google launched its Universal Commerce Protocol in January 2026, with checkout inside AI Mode and Gemini and a Business Agent in Search. Microsoft launched Copilot Checkout the same month. ChatGPT ads went from launch in February to 89 countries and more than 100,000 advertisers by October. The payment networks are following: Visa is running live agent purchases in Europe with more than 30 card issuers.
Third, retailers are splitting into open and closed. Walmart put its Sparky assistant inside ChatGPT and Gemini. Amazon keeps outside agents out, and blocked Meta’s shopping agent within two weeks of its launch, while Shopify opened its checkout to the same agent. Which side a retailer picks will decide where its brands can be found.
What will happen in 2027
2027 is the year the buying decision moves into the assistant. Forrester puts one in twenty US and UK consumers replacing traditional shopping with conversations during the year. Five things will happen, and each comes with a step a brand should take.
1. Every major assistant will have its own business console. Google has already added AI share-of-voice insights to Merchant Center. Microsoft, Perplexity, Meta and Amazon will offer brands the same claim, feed and insights tools that ChatGPT now has. What to do: name one owner for all your AI storefronts and keep a single source of truth for product data.
2. The decision gets made in the chat. Business Agents will expand beyond the US and learn to do more, such as check an order or book an appointment. Shoppers will settle their last doubts with a brand’s agent and choose without leaving the conversation. What to do: launch your agent early and test it against the hardest questions buyers ask.
3. Checkout returns to the chat, and this time brands say yes. What failed in 2026 was a single-item checkout that OpenAI ran, with no baskets, loyalty or promotions. What returns is the brand’s own checkout, opened inside the conversation. OpenAI is already working with Visa on payments for agents, and Google has checkout live in AI Mode. Brands that refused in 2026 will accept, because the sales are too large to leave to rivals. What to do: get prices, stock and policies exact, since an agent will act on whatever you publish.
4. Feed quality becomes a ranking fight. As more brands upload feeds, the complete and accurate ones will be picked first. Feed work will get its own budget, as search feeds did a decade ago. What to do: treat every product attribute as a reason to be chosen, and fill them all.
5. Paid and earned get measured together. Sponsored agents and AI ads will scale fast. They sit beside the answer and do not change it, so brands will need to know which visits came from an earned recommendation and which from a paid placement next to it. Each platform’s own numbers will stay partial, as Business Hub’s are today. What to do: set up tracking across all assistants now, so you have a baseline before the ad money arrives.
2028 to 2030: commerce when the agent is the customer
By 2030, a large share of buying will be done by AI agents acting for people. How large is an open question. The US forecasts below differ by about five times, which tells you how uncertain this is. Even the lowest, from Morgan Stanley, puts at least one in ten US ecommerce dollars through agents.
Who | Forecast for 2030 |
|---|---|
Up to $1 trillion of US sales through agents, $3 trillion to $5 trillion worldwide | |
$300 billion to $500 billion in the US, 15% to 25% of online retail | |
$190 billion to $385 billion in the US, 10% to 20% of ecommerce | |
One in ten consumers routinely buying through agents |
Here is how we expect those years to play out.
Prediction 1: Shoppers send agents. People will give their assistant standing instructions: a budget, favourite brands, things to avoid. Routine buying goes first, including groceries, household goods, travel and renewals.
Prediction 2: Agents sell to agents. The shopper’s agent will talk to the brand’s agent about price, bundles and delivery. The brand agent you launch in 2027 becomes a sales channel in its own right.
Prediction 3: Control settles at the final surface. The LLM will hold the customer’s preferences, purchase history and payment details. Brands will compete to be the default choice in that memory. Retailers without an agent of their own risk becoming fulfilment suppliers to the platforms that do have one.
Prediction 4: Rules arrive for paid placement. Regulators in the US and EU are already looking at how AI answers get shaped. We expect clear disclosure rules for paid placement around AI answers before 2030. That will raise the value of recommendations that are earned.
Nobody can draw the final map yet. The direction is the same in every signal in this article: more of the decision, the transaction and the relationship will sit with the assistant.
What the new surface takes away from brands
A brand with its own site could count on meeting the shopper there. That is the first thing to go.
Go back to the parent and the stroller. The answer named three strollers. The parent never saw a homepage, a hero video or a tagline. Each brand was introduced in the assistant’s words, and the assistant took those words from product pages, reviews and whatever else it read.
The second loss is the shelf space you could buy. A search page had ten links and a row of ads, so a brand in sixth place still got seen. The stroller answer has three names and no page two. Ads will sit beside it, labelled as ads, but OpenAI has said that ad spend does not change what ChatGPT recommends.
The third loss is the visit, and everything a visit used to give you. A shopper browsing your site could be shown a bundle, asked for an email address and reminded a week later. A shopper whose questions are answered in the chat does none of that. Business Hub gives you totals in return: how many mentions, how many clicks. It does not show the conversation, so you cannot see why the parent picked the other stroller.
What the assistant cannot take is the product, the price, the service and what customers say afterwards. Those are the things it judges you on.
The core ingredients of a winning brand
Why did those three strollers get named, and not the dozens of others on the market? Five things decide it.
A clear answer to “best for whom”. The parent asked for light, cabin-sized and under $400. The brands that say exactly that, in plain words, get matched. A stroller described as “perfect for every family” matches nothing.
Facts a machine can read. Weight, folded size, price and stock have to be on the page and in the feed as data. Adobe found that 39% of the content on retailer homepages cannot be read by LLMs at all. Shopify reports that products in its structured catalog convert twice as well as those found through general AI search.
Proof from other people. The assistant checks your claims against reviews, parent forums, YouTube tests and press. If nobody outside your own site says the stroller fits in the bin, your saying so counts for little.
A brand agent that can close. When the parent taps through and asks whether it will fit on a budget airline, the agent needs a straight answer and the dimensions to back it up. A vague reply sends the parent to the next brand on the list.
Testing before publishing. Research that Bodhium Labs co-founder Kartik Hosanagar presented in March 2026 showed that small, truthful wording changes in a product description can change whether an AI recommends the brand, most of all for brands that are not already in first place. Think of the difference between “lightweight and compact” and “weighs 6 kg and folds to cabin-bag size”. Nobody can guess which wording wins. You have to test it.
The capabilities brands still need to own
The assistant becomes the storefront. Most of the work behind the storefront stays with the brand, and some of it gets harder.
Product data used to be an IT chore. Now it is a sales channel, and someone has to keep it correct every day, the way a store manager keeps a shelf. The same goes for content and reputation. Your pages, and the outside sources that talk about you, are what assistants read, so you need people who can publish, correct and respond quickly, including when a source gets your brand wrong.
Two capabilities are new. A brand agent has to be trained, tested and watched like a new sales hire. And measurement has to cover every platform at once, because each one will show you only its own partial numbers.
Delivery, service and loyalty remain yours. With fewer site visits, the customers you know by name are worth more.
The budget moves the same way. Less of it goes to buying attention, and more goes to earning the recommendation.
Does the brand’s own UI still matter?
Yes, but it has a different job, and it now has two kinds of visitor.
The first visitor is a machine. Assistants read your pages to decide what to say about you, and they do not care how the page looks. They need clear product pages, full specs, plain policies and quick access for AI crawlers. Kartik Hosanagar calls this two front doors: one for people and one built for agents, because “a website is not the right front door for an agent”.
The second visitor is a person who has nearly decided. The parent who clicks through from ChatGPT already knows which stroller they want. The page has one job, which is to confirm what the assistant said and take the payment quickly. If the price differs from the one in the chat, you lose a sale that was already won. Walmart’s numbers show how much this last step matters: shoppers converted three times better on its own site than inside the chat.
What matters less is the part most brands spend the most on. Fewer shoppers will land on the homepage and browse. Move that money to product page accuracy, speed, checkout and the signed-in experience, where accounts, orders and loyalty live.
There is also a new interface to design. For many shoppers, the first thing they meet from your brand will be your agent’s conversation.
How a brand should position itself in this new world
It starts with a choice. In a keynote this year, Kartik Hosanagar described two ways a brand can play. One is to win the agent: be the best documented, easiest choice for a machine to make. The other is to give people an experience they seek out themselves, so they never ask an assistant to choose. Most brands will need some of both, but they should know which one leads.
Whichever leads, three rules hold.
Be specific. People ask assistants narrow questions. Own a set of them fully and stop chasing the whole category.
Cover what the shopper did not say. The parent named weight, size and price. They said nothing about the warranty, the fabric or the wheels, so the assistant decides those points for them. The brand that publishes clear facts on the unasked questions wins the tie.
Be asked for by name. This is the strongest position there is. When a shopper says “order my usual” and names your brand, no comparison happens. Brand preference becomes an instruction to the agent, so keep building demand in the real world through product quality and word of mouth.
All of this applies to B2B as well. A buyer shortlisting software or suppliers now asks an assistant first, and the same rules decide who makes that shortlist.
The playbook: what to do now, in 90 days and over 12 months
Start with the free steps this month, then build the habits that compound.
In the next 30 days
Claim and verify your business in ChatGPT Business Hub.
Fill in every profile field and upload a product feed with correct prices, stock, images and links.
Record your starting point. For your top buying questions, note where you appear today on ChatGPT, Gemini, Google AI Mode, Perplexity, Claude and Copilot.
In the next 90 days
Check that your product facts match across your site, your feed and every retailer listing.
Find the sources assistants cite in your category, and list where your brand is missing or described wrongly.
Rewrite your most important product and comparison pages so they answer buyer questions directly. Test each draft before it goes live.
If you are eligible, launch a Business Agent and test it with the hardest questions your sales team hears.
Over the next 12 months
Give feeds, content and reputation a named owner and a weekly rhythm, the way you run paid media today.
Build a program for third-party content: reviewers, affiliates, creators and communities.
Tie AI visibility to revenue. Compare sales for the products and questions where your visibility rose against those where it did not, and move budget to what worked.
Get ready for checkout in the chat: exact inventory, clear policies and payment options that agents can use.
How Bodhium Labs helps brands win this shift
Bodhium Labs is an AI lab built for this problem: getting brands recommended by AI assistants, and proving the result in revenue. The lab is led by Kartik Hosanagar, Krishna Shrinivasan and Phalgun Kompalli, and its methods come from research into how AI models choose what to recommend.
Bodhium Labs helps brands navigate ChatGPT Business Hub and every AI commerce channel that follows it. That starts with the hub itself: claiming and verifying the profile, building a product feed that gets picked, setting up and testing the Business Agent, and reading the Insights numbers correctly.
Business Hub then shows a brand its score on one platform. The Bodhium Labs AEO platform covers what the hub leaves out:
Visibility on every AI platform. See how you appear on ChatGPT, Gemini, Google AI Mode, Perplexity, Claude, Copilot, Grok, Meta AI and more, question by question, next to your competitors.
The reasons behind the score. See which sources assistants cite, how Reddit, YouTube and review sites shape the answer, and which sources get your brand wrong.
Shopping feed performance. Track how your products show up in AI shopping results, and improve the feeds behind them.
Ranked opportunities. Get a sorted list of the changes most likely to lift your visibility, so your team knows what to do first.
Testing before you publish. Bodhium Labs simulates how AI platforms will respond to a page or an article before it goes live. You compare drafts and strategies, publish the one that wins, and avoid the one that would have failed. Bodhium Labs is the only AEO company that can test content before deployment.
Content that earns the recommendation. Writer agents create web pages, articles, product pages and third-party content, each scored and improved before release.
Proof of impact. Track AI bot traffic and connect visibility gains to business results.
Bodhium Labs works with consumer brands, B2B companies and agencies.
To see where your brand stands on AI platforms today and what it would take to win, contact Bodhium Labs.
Frequently asked questions
What is ChatGPT Business Hub?
Business Hub is the dashboard OpenAI launched in beta on November 10, 2026. A business uses it to claim and verify its profile in ChatGPT, edit its details, add a product feed, and see how often ChatGPT mentions it and sends clicks to its website.
Does claiming my business make ChatGPT recommend me more often?
No. OpenAI says claiming a profile does not guarantee better placement or more sales, and ChatGPT keeps making its own recommendations. Claiming gives you correct information and a view of your numbers. The recommendation still depends on your content, your product data and what other sources say about you.
What can Business Hub not tell me?
It does not show which questions you appear for, how you compare with competitors, or how you perform on other AI platforms such as Gemini, Perplexity or Claude. For that you need a platform that tracks all of them, such as the Bodhium Labs AEO platform.
Can shoppers buy from my brand inside ChatGPT?
Not through Business Hub today. Business Agents can answer questions and link to your site, but they cannot take payment. OpenAI pulled back its first in-chat checkout in March 2026 and now sends the purchase to the merchant. Checkout will return to the chat in 2027 as brands grow more open to it.
What should a brand do first?
Claim and verify the profile, upload an accurate product feed, and record where the brand appears today across all major AI platforms. Then fix the gaps: product facts, the content assistants read, and the third-party sources they cite.
Can Bodhium Labs help with ChatGPT Business Hub?
Yes. Bodhium Labs helps brands claim and set up their Business Hub profile, build and improve the product feed, set up and test the Business Agent, and track results. It then does the same across Gemini, Google AI Mode, Perplexity, Claude and other AI platforms, so a brand can see and improve its visibility and sales everywhere shoppers ask.
About the author

Rahul Vijaywargiya
Product Manager
Product @ Bodhium Labs. Prev. MBA @ Wharton, Flipkart, BITS Pilani.