Latest

WebMCP: What is it and what do retailers need to do?

  • WebMCP is a new browser-level web standard, jointly developed by Google and Microsoft

  • AI agents are already browsing the web on behalf of consumers via products like Chrome Auto Browse, OpenAI’s Atlas, and Perplexity’s Comet

  • The implementation barrier is low and websites with clean HTML forms can become agent-ready by adding just two attributes

  • AI visibility and WebMCP are complementary, not separate

  • The time to act is now, not when the standard is finalised

There is a moment in every major technological transition where the technical foundations get laid before most people have noticed a shift is happening. Structured data before rich snippets went mainstream, mobile-first indexing before smartphone traffic overtook desktop and schema markup before AI Overviews.

WebMCP is set to be one of those moments. And if you work in retail ecommerce, it’s worth paying attention. WebMCP is a new browser-level web standard, jointly developed by Google and Microsoft, that lets websites declare their functionality as structured, callable tools for AI agents, replacing slow and unreliable screen-scraping behaviour.

So, what actually is WebMCP?

WebMCP, or Web Model Context Protocol, is a proposed browser-level web standard that allows websites to declare their capabilities as structured, callable tools for AI agents. It sits between your existing website and AI agents as a structured bridge layer, rather than requiring you to build or maintain a separate API.

But what does that mean in simple terms? Right now, when an AI agent tries to interact with a website, it essentially does what a visually impaired user might do without assistive technology; it takes screenshots, parses raw HTML and makes its best guess on which button does what.

With WebMCP, your website instead says something like: “Here is a function called searchProducts. It needs a query term and a category filter. Call it and I will give you structured results.” The agent calls the function, gets the data and moves on.

By defining these tools, you tell agents how and where to interact with your site, whether that is booking a service, navigating a product catalogue, or completing a checkout flow. This direct communication channel eliminates ambiguity and allows for faster, more reliable agent workflows.

This is a joint effort from Google’s Chrome team and Microsoft’s Edge team and incubated through the W3C, which means this is not a startup experiment. It is infrastructure being built by the people who run the browser ecosystem.

Why should retail and ecommerce teams care?

AI agents are no longer a future concept. In January 2026, Google shipped Chrome Auto Browse, powered by Gemini, OpenAI’s Atlas browser launched with Agent Mode and Perplexity’s Comet is doing full-task browsing across platforms. These are products being used to some extent by real users and the number of people delegating shopping research and purchasing tasks to AI agents is still developing, but growing quickly.

The question for any retailer is no longer “will AI agents interact with our site?” It’s “when they do, will they be able to complete the task, or will they bounce to a competitor who made it easier?”

This is the responsive design moment for agentic search. When mobile arrived, the sites that adopted responsive design early won the distribution game. The late movers scrambled to catch up while traffic shifted. The same dynamic is playing out now with agent-readiness. The retailers who get ahead of this will have an advantage as leaders as agentic commerce becomes mainstream.

For ecommerce specifically, the implications are direct. Users can better shop your products when agents can easily find what they are looking for, configure shopping options and navigate checkout flows with ease and precision. When an agent can call a structured search products tool with filters and receive clean results, it will always prefer that over scraping a category page and trying to interpret the DOM.

How does it actually work?

WebMCP gives developers two ways to make websites agent-ready: a Declarative API and an Imperative API, but what is the difference between these?

The Declarative API

This is the quick win option and the one most relevant for retailers in the short term. If your site already uses standard HTML forms, you can make them agent-compatible by adding just two attributes: “toolname”, which names the tool, and “tooldescription”, which tells the agent what the tool does. Forms and features such as search forms, filter panels, add-to-basket flow and checkout forms can become structured tools for AI agents with minimal development work.

There is also an optional “toolautosubmit” attribute, which makes the form submit automatically when the agent fills it in.

The Imperative API

This is for more dynamic interactions. Developers register tools programmatically through a new browser interface called “navigator.modelContext”. Tools can be registered and unregistered based on page state, so a checkout tool only appears when items are in the basket and a booking tool surfaces only after dates are selected. The agent only ever sees what is contextually relevant, which makes interactions faster and more reliable.

The emerging best practice is that every form should be declared as a WebMCP tool. The implementation cost is very low and the alternative is not that agents will not use your site, but they will just take longer to complete actions and be more likely to fail.

What this means for your SEO and content strategy

It is important to be clear about where WebMCP sits relative to the broader AI visibility picture because the two are related but distinct.

The AI visibility work you are already (hopefully) doing, getting your brand cited in ChatGPT, Perplexity, Google AI Overviews, building entity authority and structured data, remains entirely valid and increasingly urgent. That visibility is what gets an AI agent to your site in the first place.

WebMCP adds a layer beyond content retrieval. It is about making your website’s functionality accessible to AI agents. Not just “can an LLM find and recommend my product?” but “can an AI agent actually complete a purchase on my site?”

Think of it as two complementary problems. Content and authority work gets you into the consideration set. WebMCP implementation determines whether you can convert that consideration into a completed transaction, even when there is no human clicking the buttons.

The brands already visible in AI answers are the ones agents will route users to first. Getting mentioned and cited today is how you get to the front of the queue for agentic AI actions in the future.

What should you actually do right now?

WebMCP is still in early preview and the spec will evolve. But the actions below are worth taking regardless of how quickly the standard matures.

Audit your transactional touchpoints

Identify the five to ten most important actions on your site: product search, filter and sort, add to basket, checkout, account creation, click-and-collect booking and returns initiation. For each one, ask whether the underlying form or interaction is built on clean, semantic HTML with clear labels and predictable inputs. If the answer is no, then that’s a technical debt problem regardless of WebMCP, because it also affects accessibility and crawlability.

Prioritise your search and filter functionality

For retail, product search is the highest-value WebMCP tool you can implement. An agent that can call a structured product search with category, price range and availability filters and get clean results will strongly favour your site over one it has to scrape. This should be your first conversation with your development team.

Get serious about form hygiene

If your website has clean, well-structured HTML forms, you are most of the way to WebMCP readiness already. The barrier is not the two new attributes, it is years of forms built as JavaScript workarounds, inconsistent labels, unstable redirects and inputs that only a human could interpret contextually.

Start the developer conversation now

Even if full implementation is a year away, the teams that start experimenting now will move faster when the standard lands. Share the Chrome early preview documentation with your developers and frame it as awareness-building rather than an immediate sprint. You can enable WebMCP for testing in Chrome 146 via chrome://flags/#enable-webmcp-testing, and Google has published a live travel demo showing the full tool discovery and invocation flow.

Brief your clients or stakeholders

If you are agency-side, this is a conversation worth having proactively. Frame it the right way, not as something requiring urgent investment today, but as a signal of where the web is heading and evidence that the technical fundamentals you have been recommending, clean markup, structured data, semantic HTML etc, were always the right foundations.

The bottom line

The web is going to be rebuilt to serve two types of users: humans and AI agents. WebMCP is how Google and Microsoft are giving agents a native, structured way to interact with websites, without the fragility of screen scraping or the overhead of building and maintaining separate APIs.

For retailers, the stakes are straightforward. If an AI agent is helping a consumer find and purchase a product and your site is harder for that agent to use than a competitor’s, you will lose that transaction. You will not necessarily know you lost it, because there will be no session recorded, no bounce rate logged, no visible signal. The agent will simply route the user elsewhere.

The good news is that the foundations required, clean HTML, semantic forms, structured data and clear transactional flows, are the same foundations that have always underpinned strong technical SEO. You are not starting from scratch, you should be building on work that was already the right work to be doing.

Do not wait until WebMCP is fully standardised across all browsers to start paying attention. By the time it is obvious, the advantage will already belong to your competitors.

For more information on how we can help you when it comes to all things SEO, get in touch with the team today.

 

Ready for change? Let's talk

Speak to Summit