If you’ve spent any time on this website, you know I’m a firm believer that structured data is the foundation of the modern web. For years now, we’ve used Schema.org to tell Google what our content is about. We’ve marked up our products, our prices, and our local businesses so that search engines could display those nice rich snippets we all strive for.
But as we move into 2026, the game is changing. We aren’t just optimising for a “Search Engine” anymore, we’re optimising for AI Agents like Gemini, ChatGPT and Claude.
These agents, as well as the new agentic layer in Chrome, don’t just want to read your website anymore, they want to use it. To make that happen, we need more than just static labels. We need a way for agents to interact with our code.
That’s where WebMCP and UCP come in. Keep reading to learn how these three technologies work together to turn your e-commerce site into a high-performance machine for AI agents.
Here’s an image describing basically how it works:
Structured Data – The What
Structured data remains the main foundation. Before an AI agent can buy a product or book a service, it has to find it.
When you markup your product with Product, GTIN, Offer and AggregateRating, you’re providing the concrete facts that AI Agents need to find your products. It’s almost like your product having its own ID card.
In the agentic SEO era, if your Schema markup is missing the shippingDetails or hasMerchantReturnPolicy, an agent might skip your store entirely because it can’t give the user a guaranteed “total price.”
The Role: Structured data provides the vocabulary. It tells the agent: “This is a ‘Black Leather Boot’, it costs £120, and it is ‘In Stock’.”
WebMCP – The Hands
This is the newest piece of the puzzle. The Web Model Context Protocol (WebMCP) is a browser-level API that gives your website “hands.”
Up until now, AI agents had to “scrape” your website, literally guessing which button was the “Add to Cart” button. WebMCP replaces that guesswork with something called a Tool Contract. By adding simple HTML attributes like toolname and tooldescription to your forms, you are telling the agent: “Don’t guess. Use this specific function to check my stock or add this item to the basket.”
The Role: WebMCP provides the capability. It tells the agent: “Here is the tool you need to call to select the ‘Size 6’ variant and check if it’s available for next-day delivery.”
UCP – The Wallet
Finally, we have the Universal Commerce Protocol (UCP). If structured data is the “What” and WebMCP is the “Hands”, then UCP is the “Wallet”.
UCP is an open standard (co-developed by Google and Shopify) that allows the entire checkout process to happen inside the AI interface. Instead of the agent sending the user to your checkout page (where they might get distracted and drop off), UCP allows the agent to pull your real-time tax, shipping, and payment logic via a standard manifest file located on your server at /.well-known/ucp.json.
The Role: UCP provides the completion. It handles the payment and order creation directly, ensuring a frictionless sale while you remain actual store it purchases the item from.
Putting It All Together
Imagine a user asks their AI assistant: “I need some durable black leather boots around £120 in a size 6.“
-
Discovery (structured data): Gemini then scans the web and finds your product page offering black leather boots. Because you’ve implemented brilliant bespoke structured data markup (Product, Offer, Color, Material), Gemini can instantly verify the price and material. It doesn’t have to “guess” whether or not they are leather, how much they are or whether or not they are coloured black – your structured data confirms it.
-
Interaction (WebMCP): The customer likes the look of the boots so they ask Gemini “Are these in stock for delivery tomorrow?”. Gemini then uses your WebMCP Tool Contract (linked to your inventory search). Your site “replies” through the protocol with real-time stock levels and delivery times. The agent tells the user: “Yes, there are 3 pairs left and they offer next-day delivery”. This turns your website from a static page into a live assistant.
-
Transaction (UCP): The customer tells Gemini: “Perfect, buy them for me”. Gemini checks your
/.well-known/ucp.jsonmanifest to see how to handle the payment, and finds the “Buy Now” button. The user clicks it, and UCP handles the secure payment through Google Pay, instantly triggering the Shopify “Order Complete” email and the product has been ordered by the customer.
This isn’t just a search anymore! It’s a multi-step negotiation between the AI and your website. The customer never had to leave the AI.
Does This Mean SEO is Dead?
Traditional SEO isn’t dying – it’s evolving into Agentic SEO.
If you want your e-commerce site to thrive, you can no longer afford to be invisible to AI agents – you must be visible to them in the first place! Yes, it means less human traffic to your website, but a combination of structured data, WebMCP and UCP could result in more sales, which is ultimately what you want from your website in the first place.
Where Do I Start?
Start with robust structured data and a strong brand entity, then move onto WebMCP for your site’s products, and get your UCP manifest ready for the checkout process.
