Google just dropped something big, and almost nobody in the local business world is talking about it yet.
Chrome 146, now available in early preview, includes a new standard called WebMCP (Web Model Context Protocol). In plain terms, it lets AI agents interact directly with a website's functionality instead of trying to "browse" it the way a human would. Think of it as giving AI a direct line to your business instead of forcing it to squint at your homepage and guess where the booking button is.
If you've been following the shift toward AI-powered search, this is the moment where things get real. WebMCP isn't just a developer tool. It's the foundation of how AI agents will discover, evaluate, and recommend businesses in the very near future.
And if your website isn't ready, you're going to be invisible to a growing slice of potential customers.
What Actually Happened
On February 10, 2026, Google's Chrome team officially published the WebMCP early preview, a joint effort between Google and Microsoft. The announcement was technical, aimed at developers, but the implications reach far beyond code.
Here's the short version: right now, AI agents (like the ones powering ChatGPT, Google's Gemini, Claude, and other tools) interact with websites the way a blindfolded person navigates a room. They load the page, try to read the text and layout, click buttons, fill out forms, and hope for the best. It's slow, fragile, and unreliable. That's why AI-powered search results sometimes get details wrong or miss information entirely.
WebMCP changes that. It lets website owners define structured "tools" that tell AI agents exactly what actions are available on a page, like searching for a menu item, checking appointment availability, or submitting a contact form. Instead of guessing, the AI can just ask the website directly: "What can I do here?" and get a clean, structured answer.
There are two ways it works. Developers can register tools through JavaScript using a new browser API called navigator.modelContext, which gives fine-grained control over what's exposed. Or, for simpler use cases, existing HTML forms can be tagged with special attributes that make them agent-readable, no custom code required.
Why This Matters for Search and Discovery
For the past 25 years, search has been a one-way street. You build a website, optimize it for Google's crawlers, and hope the algorithm ranks you well. The customer types a query, scrolls through results, clicks a link, and figures out what to do from there.
That model is already cracking. AI Overviews, ChatGPT search, Perplexity, and voice assistants are changing how people find information. But the underlying mechanism has been the same: AI reads your website content, summarizes it, and presents it. WebMCP introduces something fundamentally different.
Instead of just reading your content, AI agents will soon be able to take action on your website. Not browse it. Not screenshot it. Actually use it, the way a knowledgeable human assistant would.
Industry experts are already calling this the shift from the "searchable web" to the "agentic web." One analysis framed it this way: websites will soon need two layers. A human-focused layer that's visual, branded, and narrative-driven. And an agent-focused layer that's structured, schema-based, and fast. The businesses that build both layers win. The ones that only have the first layer start losing ground.
This isn't hypothetical. The W3C (the organization that governs web standards) is already incubating WebMCP through its Web Machine Learning Community Group. Google and Microsoft are building it into their browsers. The trajectory here is as clear as mobile-responsive design was in 2012.
What This Means for Local Businesses
Let's bring this home. If you run a restaurant, a plumbing company, a law firm, a yoga studio, or any local service business, here's what's coming.
Today, when someone asks an AI assistant "find me a plumber in Medford who can come this week," the AI searches the web, reads some websites, maybe pulls from Google Business Profiles, and gives a recommendation based on what it can piece together. If your website is slow, cluttered, or missing key information, the AI might skip you entirely.
With WebMCP, the picture changes. A plumbing company's website could expose tools like "check_availability" or "request_estimate" that an AI agent can call directly. A restaurant could expose "view_menu," "check_hours," or "make_reservation." A law firm could offer "schedule_consultation" or "check_practice_areas."
The AI doesn't have to guess. It doesn't have to scrape your homepage and hope it finds the right phone number buried in your footer. It gets clean, structured access to exactly what your business offers and can act on it instantly.
This is the natural next step after schema markup and structured data. We've spent years helping businesses tell search engines what they are. WebMCP lets businesses tell AI agents what they can do.
What This Looks Like in Practice
Imagine a customer in Boston asks their AI assistant: "Book me a table for four tonight at an Italian restaurant in the North End with outdoor seating."
Without WebMCP, the AI searches, finds some restaurants, reads their websites, and gives a list with phone numbers. The customer still has to call.
With WebMCP, the AI queries each restaurant's exposed tools directly: Does this restaurant have availability tonight? Do they have outdoor seating? Can I book a table for four? The AI finds the match and makes the reservation, all in one step. The restaurant that's WebMCP-ready gets the booking. The one that isn't doesn't even make the shortlist.
That's the competitive advantage we're talking about. And it's not limited to restaurants. Home service businesses with online scheduling, professional services with intake forms, fitness studios with class bookings: every business that lets customers take action online is a candidate for this.
What You Should Be Doing Right Now
WebMCP is in early preview, which means it's not mainstream yet. But "not mainstream yet" is exactly where the smart money moves. Here's what we recommend for local businesses that want to stay ahead.
1. Get Your Structured Data Right
This is table stakes, and it's where most local businesses are still falling short. Schema markup, Google Business Profile optimization, and clean, consistent NAP (name, address, phone) data across all platforms: these are the building blocks that AI agents already rely on. WebMCP will amplify the importance of structured data, not replace it. If your foundation is weak, nothing else matters.
2. Make Your Website Action-Oriented
Look at your website through the lens of "what can someone do here?" Not what can they read, but what can they do. Can they book an appointment? Request a quote? Check availability? View a menu? If the answer is "they can call us," you're behind. Every key action should be available through a form or interactive element on your site, because those are the exact touchpoints WebMCP will expose to AI agents.
3. Clean Up Your Site Architecture
AI agents, with or without WebMCP, reward clean structure. That means clear page hierarchies, logical URL structures, fast load times, and content organized by intent. If your website is a maze of outdated pages, broken links, and inconsistent formatting, fix that before you worry about anything else.
4. Invest in Schema Markup Now
If you're a local business and you don't have LocalBusiness schema, Service schema, FAQ schema, and (where applicable) Menu or Event schema on your website, you're leaving money on the table today and setting yourself up to be invisible tomorrow. Schema is the language AI agents already speak. WebMCP extends that language into actions.
5. Watch This Space Closely
WebMCP is early. The standard is still evolving, and it'll be months before mainstream adoption. But the direction is clear, and the businesses that prepare now will have a significant head start. Follow Google's Chrome developer blog, keep an eye on the W3C Web Machine Learning Community Group, and partner with a marketing team that understands where AI search is headed.
The Bigger Picture: The Agentic Web Is Here
We've been talking about AI search optimization for a while now, explaining to our clients why Answer Engine Optimization matters, why query fan-out changes the game, why structured data is more important than ever. WebMCP is the biggest validation yet that this shift is real and accelerating.
The phrase we keep coming back to: we don't optimize for just one search engine anymore. We build systems for wherever search happens. Google, Bing, ChatGPT, Perplexity, voice assistants, and now AI agents that interact directly with your website.
The web is becoming agent-ready. The question for your business isn't whether to adapt. It's whether you want to be early or late.
We know where we'd rather be.
