Prepare for WebMCP: Why Now Is the Time

New technologies emerge constantly, and I learned early in my career that chasing every trend is a recipe for wasted time. After burning hours on fleeting innovations like Google Authorship, I realized that waiting for broader adoption, learning from early adopters’ mistakes, and moving quickly when the time is right creates far more value.

But every so often, a shift arrives where early movers don’t just win in the current landscape,they define the next one. Think of the first SEOs who read the PageRank paper and realized building links mattered. WebMCP feels like that moment, only bigger.

This isn’t just another evolution in search or generative engine visibility. The very location where discovery happens is changing, and so is the entity doing the discovering.

Non-human engagement is coming. For years, SEOs debated optimizing for search engines versus humans (the answer, unsurprisingly, is both). That debate is about to be upended. What happens when discovery shifts from a person to a large language model or an agentic system?

It’s already happening. When you ask ChatGPT something, it runs searches, asks follow-ups, and returns conclusions. The agent plans and decides for you, shaped entirely by what it retrieves and how it interprets it. You can see these supplemental queries in DevTools.

I see this as the latest chapter in a longer story:

• Discovery v1: People experienced the world firsthand. Word of mouth was the discovery point.I believe Discovery v5 will be the most dramatic shift since v2. Imagine offloading basic decisions to agents, freeing you for more important things. I’ve seen this utopia before, but the world we’re creating is fundamentally different for marketers. WebMCP is one of the first concrete steps on this journey.The trust ratchet only turns one way. Do you accept AI Overviews more often now than when they launched? Not always, but more often. For quick, low-risk queries, trust is easy. As these systems improve, that trust boundary moves. What we let agents do on our behalf expands.The cost of automating a grocery reorder is small. The benefit of an agent monitoring flight deals within your budget is high. The appeal of an autonomous vehicle taking your family to Disney World while you sleep is compelling. You may say you’ll never hand over autonomy. People said the same about search engines, smartphones, and GPS. The path is predictable: skepticism, reluctant adoption, then dependency.What does this have to do with WebMCP? MCP servers and skills files are early infrastructure for Discovery v5, but they’re high-barrier and context-specific. WebMCP is different. It’s a browser-native web standard, currently a W3C Community Group Draft and in early preview in Chrome 146 beta. It lets websites expose actions to AI agents without scraping or brittle automation.This isn’t Google-only. The spec is co-authored by engineers from Google and Microsoft. When two of the largest browser and AI vendors write the spec together, it has a different trajectory than a unilateral bet.Right now, when an AI agent tries to act on your site,filling a form, booking an assessment,it reads your DOM, guesses field meanings, and hopes the date format matches. It’s intelligent but fragile. One UI change breaks the flow.WebMCP changes this by letting you tell the agent exactly what your site can do and how. The spec defines two approaches: declarative and imperative.Declarative vs. imperative: You already know this distinction. The Declarative API is the one to act on now. You annotate existing HTML forms with attributes describing what each field means. The browser translates this into a structured tool agents can call. The form works normally for humans. The agent gets a clean interface.Think of it like schema markup in its early days: the syntax evolved, but the idea of annotating what exists for machine understanding was clear. You’re not building a new system. You’re making what you have legible to a new visitor type.The Imperative API is more mature in the spec and available for testing. You register tools directly in JavaScript. For example, a site taking bookings can register a `book-assessment` tool with input schema and execution logic. This is more powerful for dynamic or multi-step flows. Tools can change based on state, like a hotel booking demo where searching unlocks filtering and booking tools.Start with declarative for low lift and high legibility. Extend into imperative as needs grow.What the agent sees: Before and after. Take a standard booking form. Without WebMCP, an agent guesses what each field means. With declarative annotations, you add `toolname`, `tooldescription`, and `toolparamdescription` attributes. The form looks and works the same for humans. But the agent now understands exactly what the form does and how to use it.Proposed attributes include `toolname`, `tooldescription`, `toolautosubmit` (for agent submission without consent), and `toolparamdescription`. The declarative API is still being specified, but the concept is settled.Why this matters for your sites. Think about queries agents will handle in Discovery v5: finding an SEO consultant, comparing tools, or planning an event. Which site gets the engagement? The one the agent can interact with cleanly. If your competitor has WebMCP-registered tools and you don’t, the agent completes the action on their site. The user may never know they had a choice.Tool descriptions are the new meta descriptions. Quality names and parameters shape whether an agent selects your tool. The best practices guidance reads like conversion copywriting: use clear verbs, explain the why, and be specific. You’ve been writing for machine readers for years. This is the next layer.The window is open, but not forever. I’ve been skeptical of early adoption my whole career. But I recognize moments that are different in kind, not just degree. Schema markup, SSL, and mobile optimization were such moments. Early movers earned disproportionate returns. The people who understood the underlying shift compounded that advantage.WebMCP is a W3C Community Group Draft today, co-authored by Google and Microsoft, running in Chrome 146 beta, and integrated into Cloudflare’s infrastructure. It’s not table stakes yet. But the trajectory is clear: the spec matures, browsers ship it, agents prefer sites with structured tools, and those without become invisible to this visitor class.The declarative approach, once finalized, means starting is genuinely low-effort: annotations on your most important forms. The imperative API is available for testing right now.That’s why I’m making this argument now, not in six to 12 months when everyone else is trying to catch up.