You don't have to wait: agentic buying already works today — inside the shopper's own AI

There's one question every merchant asks the moment agentic commerce comes up: "When is this actually usable?" And underneath it hides an assumption — that some big platform has to switch it on first. That we're waiting on ChatGPT, on Gemini, on the commerce platform. That agentic buying is a queue you stand in.

That assumption is wrong. You aren't waiting on anyone.

The same purchase already works today — via a connector

There are two roads to agentic buying. One is the road the press releases talk about: a platform indexes your store, and one day a shopper's AI surfaces you by chance as a recommendation. That road is genuinely coming — but it runs on someone else's schedule, and you don't have a date.

The other road is open right now: the shopper adds your store's connector to their own AI. One link into Claude or ChatGPT, connected once — and after that, buying is just a conversation. The shopper asks for the right product, assembles an order, and approves the payment right inside the AI they already use. No browsing, no separate portal, no waiting for a platform to notice you.

The difference is this: on the first road, you hope the AI finds you. On the second, you hand the shopper a direct pipe into your store — a pipe the shopper connects themselves and one they own.

We're not claiming it — we proved it

We built and piloted exactly this: a real AI account, with the store's connector added, made a real purchase that landed as a real order and payment — with a human approving that payment under EU strong-authentication rules (PSD2/SCA). Not a mockup. Not a demo reel. An end-to-end purchase.

See how we make it work today →

That payment ran through a real payment provider in a test environment, with a human approving every payment. It's an important line and we keep it visible: in the EU an agent never pays on its own — the shopper confirms every payment. That isn't a shortcoming of our build; it's the only legally sound way, and exactly where the market landed.

Who this is for already

Two groups get value from this immediately, without anyone "switching anything on":

Repeat customers. You share an "add this store to your AI" link — by email, as a QR code on the package, on a card in the shipment. The customer adds it once, and after that "reorder my usual" is enough. Their AI handles the rest.

B2B procurement — the strongest case. A professional buyer adds the supplier's connector to their own work AI and orders in the very tool where they already plan and work. Contract pricing and account known, no separate portal, no logins. When your customer's engineer is already speccing products in Claude or ChatGPT, the supplier whose data is already there is the preferred partner — everyone else is left competing for a conversation they weren't invited to.

A connector is only as good as its data

Here's the part that's easy to underestimate. That link, and the connector behind it — it's a pipe. But a pipe sells nothing if the data at the end of it is thin.

When a shopper's AI asks "which of these fits my case," it makes that choice from what it finds in the data. If your products carry only a one-line description and a half-empty product record, the agent can't recommend them — and can't buy them. But if your store's data layer holds far more structured, contextual information than the page itself shows — compatibilities, use cases, sizing logic, the dozens of details that decide one specific buyer's question — then the agent finds the right product and closes the sale. The connector is the third rung. The foundation is always the data. That's why the same work we do for AI visibility is also the work that makes agentic buying possible at all.

And when the platforms open up later — you're already there

That first road, platform-driven auto-discovery, is coming — roughly late 2026 into early 2027. But the timing works in your favour: the same data work and the same store connector you build today to work inside the shopper's own AI is exactly what you need when the platforms open auto-discovery. You don't build twice. You start now, you get value now from reorders and B2B, and when the queue finally moves, you're already standing at the front.

How a store's connector gets into the shopper's AI — today

Here's the whole method. We don't hide the recipe; the value is in doing it well and doing it now.

1. We build and host your store its own connector. It's a small MCP server at a stable address: a set of tools a shopper's AI can call — search the catalog, add to cart, view cart, pay, confirm the order. It talks to the store's existing API and payment provider. You switch no platform and open no new contracts.

2. We make the connection a single link. The shopper needs no technical skill: we give you an "add this store to your AI" link and two lines of instructions (in Claude: Settings → Connectors → add; in ChatGPT: Settings → Connectors/Apps → add). A one-time connection, made as smooth as the platforms allow.

3. You share the link with the people who already trust you. This is the whole insight — and why you don't have to wait. You don't have to hope ChatGPT "finds" your store, because you already have the customer relationship. You just extend it into the customer's AI. The link goes where the customer already is: the post-purchase email, a QR code on the package, a card in the shipment, a message from the B2B account manager — with, say, a discount on the first AI-placed order as an incentive. Discovery isn't luck from some platform's algorithm; it's your own distribution to your own audience.

4. The data layer is what makes it sell. The connector is a pipe; what shows through it is the store's structured, contextual data. That's why the same data work above is part of this same build — without it the connector connects but doesn't sell.

5. We keep it alive and future-proof. Connector specs are moving; we keep your server compatible as they change. And the same server plus data is already ready for the moment mainstream agents begin auto-discovering a store's MCP straight from its data (well-known standards, roughly late 2026 / early 2027). You don't build twice.

That's it — the whole picture, start to finish. And here's the point worth saying out loud: what you buy from us isn't this recipe. You just read it for free. What you buy is a built, hosted, measured and maintained system — the connector, the data layer behind it, the payment handoff and EU compliance, done for you and proven to work. Reading a recipe is easy. Building the whole stack and keeping it reliably buyable inside a shopper's AI is another thing — and it's done.

Agentic buying isn't a queue you stand in. It's a link you hand your customer. See how it works today → or book a mapping and we'll show you what your store needs.

This is the first part of a hands-on series on agentic commerce. Upcoming parts go deeper on two things this post only touched: the data layer behind the connector — why contextual structured product data, of which the data layer holds far more than the page shows, is the real moat; and the distribution playbook — how you actually get customers to add your store to their AI (post-purchase email, a QR on the package, B2B onboarding, incentives).