For as long as there has been online shopping, the basic path has been familiar: a customer searches, scans a page of results or a marketplace listing, clicks around comparing options, and eventually buys. In 2026 a genuinely new path opened up alongside it. People can now ask ChatGPT what to buy, get a recommendation, and complete the purchase without ever visiting a search page or a store the traditional way, with the AI doing the finding, comparing, and increasingly the buying. This is being called agentic commerce, and it changes who stands between your product and your customer.
For a small business that sells online, this is worth understanding calmly rather than either ignoring or panicking about, because it is early and evolving but pointed in a clear direction. The shift is not that your website disappears, it is that a powerful new intermediary, the AI assistant, is inserting itself into how people discover and choose products, and being the product it recommends is becoming its own thing to work at. This article explains what actually launched, the real shift underneath the headlines, what is genuinely in effect today versus still settling, and the practical steps to make sure AI-driven shopping works for your business rather than around it.
ChatGPT now lets people buy products directly in the chat, and a new open standard is bringing Etsy, over a million Shopify merchants, and eventually tens of millions of businesses onto AI-driven shopping. The big shift is that an AI, not a search page, increasingly decides which products a customer sees, so being the product the AI recommends is the new goal. Notably, the buying itself is settling toward discover-in-AI, buy-on-your-site, because shoppers prefer to check out where they already have accounts. The practical response for a small business is to make your product information clear, accurate, complete, and well-reviewed everywhere online, because that is what AI draws on to decide what to recommend, and to keep your own store strong since that is where the purchase still often completes.
What actually launched
In February 2026, OpenAI introduced Instant Checkout, a feature letting people buy products directly inside ChatGPT rather than being sent elsewhere to complete the purchase. It is powered by the Agentic Commerce Protocol, a new open standard developed by OpenAI together with Stripe, which is essentially a common set of rules that lets merchants and AI shopping systems connect so that products can be found and bought through the AI. The significance of an open standard is that it is not limited to one company, so it can bring a very large number of merchants onto AI-driven shopping over time.
The scale of merchant participation is what makes this more than a curiosity. Etsy sellers were live early, over a million Shopify merchants are being brought on, and payment providers connecting to the protocol are expected to bring tens of millions of additional businesses onto the platform. Set against ChatGPT's hundreds of millions of weekly users and a very large volume of shopping-related questions every day, this makes the AI a genuinely significant new commerce channel, not a fringe experiment. For merchants, the arrangement is designed to be low-friction: a small fee on completed purchases, no effect on the customer's price, and no influence on the AI's product results, while the merchant keeps control of payments and the customer relationship.
One important nuance emerged quickly and it matters for how you respond. Although the initial vision was to complete checkout entirely inside ChatGPT, the approach shifted through 2026 toward letting people discover and decide in the AI but complete the purchase where they already shop, because shoppers preferred finishing in environments where they had accounts, saved payment methods, and order histories. So the emerging shape is discover in AI, buy on your own site, rather than the AI swallowing the entire transaction, which is a meaningful and reassuring detail for merchants who worried about losing the customer relationship entirely.
The real shift underneath
Strip away the mechanics and the fundamental change is about who decides which products a customer sees. In traditional online shopping, the customer sees many options, a page of search results, a marketplace full of listings, and makes their own choice among them, which means being visible and competitive among many was the game. In AI-driven shopping, the customer asks the AI and often receives a small number of specific recommendations, or even one, which means the AI is doing the filtering that the customer used to do, and being among the few products it selects becomes far more important than being one option among many.
This is a genuinely different competitive dynamic, and it is the same underlying shift we described for search in our piece on zero-click search and being cited by AI, now arriving in shopping. When an intermediary AI narrows the field before the customer ever chooses, the question stops being how do I rank on a crowded page and becomes how do I become one of the few options the AI puts forward. That concentration raises the stakes of being recommended, because the gap between being the AI's pick and being invisible to it is much larger than the gap between ranking third and ranking eighth on a traditional results page.
The reassuring part, and the reason not to panic, is that the AI is not choosing randomly or purely by who pays, it is choosing based on information, on what it can find and understand about products and how well they match what the customer asked for. That means influence over whether you get recommended is largely influence over the information about your products that the AI can access and trust, which is something a small business can actually work on. The shift concentrates the competition, but it also makes the basis of that competition, clear and credible product information, something within your control rather than a mystery.
What is real now vs still settling
It helps to separate what is genuinely in effect from what is still moving, so you neither over-react nor dismiss it. What is real now is that AI-driven product discovery is happening at meaningful scale, with huge numbers of people using AI assistants to research, compare, and decide on purchases, and with major platforms and millions of merchants being connected to AI shopping. This part is not speculative, the behaviour and the infrastructure both exist today, and the number of people who start their shopping by asking an AI is substantial and growing.
What is still settling is exactly how and where the transaction completes. The early idea of buying entirely inside the AI has given way, at least for now, to a model where the AI drives discovery and decision but the purchase often finishes on the merchant's own site or usual channel, and the balance may keep shifting as the technology and shopper habits evolve. So the discovery shift is here and durable, while the checkout mechanics are still in motion, which tells you where to focus: on being discovered and recommended, which matters regardless of where the final click happens, rather than betting everything on any one checkout arrangement that is still changing.
This distinction is genuinely useful for a small business because it clarifies the durable investment. The thing that pays off no matter how the checkout mechanics land is making your products easy for AI to find, understand, and confidently recommend, and keeping your own store strong so that the purchase completes smoothly when the customer arrives to finish it. Both of those are worth doing today with confidence, because they serve you across every version of how agentic commerce ends up settling, whereas chasing the exact current checkout implementation is less wise while it is still visibly in flux.
How to get the AI to pick you
Since the AI recommends based on the information it can find and trust about your products, influencing whether it picks you comes down largely to the quality, clarity, and reach of that information. The foundation is having clear, complete, accurate product information wherever your products appear, detailed and honest descriptions, correct specifications, good images, and the concrete facts a customer might ask about, because an AI cannot confidently recommend a product it cannot clearly understand, and vague or thin product information puts you at a disadvantage against competitors whose listings are richer and clearer.
Reviews and reputation matter too, because AI shopping systems, like the customers they serve, weigh signals of trust and quality, so genuine positive reviews and a solid reputation make your products more likely to be surfaced and recommended. So does consistency and accuracy of your product information across the different places it appears, since inconsistency or errors undermine the AI's ability to represent you correctly. And being connected to the platforms and standards through which AI shopping operates, for many small businesses via the e-commerce platform they already use, is increasingly part of being eligible to be recommended at all, which is worth checking with whatever platform runs your store.
You will notice that none of this is exotic, and that is the encouraging theme. Clear and accurate product information, genuine reviews, a good reputation, and being present on the standard platforms are all things a well-run online store should be doing regardless, and they happen to be exactly what makes you visible and recommendable in AI-driven shopping. The arrival of agentic commerce has not invented a strange new discipline so much as raised the stakes on doing the fundamentals of online selling well, which means a small business that sells thoughtfully is already most of the way toward being AI-recommendable.
What to actually do
The practical priorities follow directly. First, get your product information into excellent shape, clear, complete, accurate, and honest descriptions, specifications, and images everywhere your products are listed, because this is the raw material the AI uses to decide what to recommend and it is the single highest-leverage thing you control. This work also serves your human customers and your ordinary search visibility, so it is never wasted regardless of how agentic commerce evolves, which makes it the safest place to start.
Second, keep your own store strong and your checkout smooth, because the emerging model has purchases often completing on your own site after discovery in the AI, so the experience a customer meets when they arrive to buy still matters enormously. And check with your e-commerce platform about how it connects to AI shopping standards and channels, since for many small businesses participation will come through the platform they already use rather than anything they build themselves, and being connected is part of being eligible to be recommended. Our guide to the Shopify 2026 updates for merchants covers how one major platform is wiring this in.
Third, keep perspective and do not over-rotate on any single fast-changing detail, because the discovery shift is durable while the exact checkout mechanics are still settling. Focus your effort on being discoverable and recommendable, which pays off across every version of how this lands, rather than chasing the precise current implementation. If you want help getting your product information and store genuinely ready for AI-driven shopping, and identifying where the biggest gaps are, that is exactly the kind of practical readiness our €49 audit is designed to map, alongside our broader guide to the e-commerce tasks worth automating.
The bottom line
People are starting to shop inside ChatGPT, and with an open standard bringing Etsy, over a million Shopify merchants, and eventually tens of millions of businesses onto AI-driven commerce, a genuinely new shopping channel is opening where an AI, not a search page, increasingly decides which products a customer sees. The fundamental shift is that being the product the AI recommends is becoming its own goal, because the AI narrows the field before the customer chooses, which raises the stakes of being picked while also making the basis for it, clear and credible product information, something you can actually work on.
The reassuring nuances are that the discovery shift is real and durable while the checkout mechanics are still settling toward discover-in-AI, buy-on-your-own-site, and that getting recommended comes down to fundamentals a good online store should do anyway: excellent product information, genuine reviews, a strong reputation, and presence on the standard platforms. So the right response is neither to panic nor to ignore it, but to get your product information into excellent shape, keep your own store and checkout strong, connect through your e-commerce platform, and focus on being discoverable and recommendable rather than chasing every shifting detail. Do that, and the new AI shopping channel becomes an opportunity that works for your business rather than a wave that passes it by.
Sources
- OpenAI — Buy it in ChatGPT: Instant Checkout and the Agentic Commerce Protocol
- Stripe — Stripe powers Instant Checkout in ChatGPT and releases the Agentic Commerce Protocol
- CNBC — OpenAI's first try at agentic shopping stumbled. It's trying again
- Checkout.com — OpenAI's agentic commerce shift: What it means for merchants
- Elogic Commerce — ChatGPT Commerce & Agentic Shopping Statistics 2026
- Azoma — ChatGPT Shopping: What Merchants Need to Know About Instant Checkout
- Digital Applied — Why AI Checkout Stalled: Discover in AI, Buy on Site
- Launch Commerce — Agentic Commerce in 2026: How to Make Sure AI Shopping Agents Pick Your Store