The AI-Native Web

From Screen-Scraping
to API Calls

Websites stop being pages. They become callable services for AI.

Try WebMCP Learn More

89%

Fewer Tokens

vs screen scraping

97.9%

Success Rate

task completion

53%

Lower Cost

operational savings

100%

Structured

machine-readable

The Evolution

From fragile parsing to reliable protocols

The Old Way Fragile

// Screen scraping - breaks easily const response = await fetch(url); const html = await response.text(); const doc = parser.parse(html); // Fragile - breaks when CSS changes const title = doc.querySelector('.header h1')?.textContent; const price = doc.querySelector('.price-span')?.dataset.value; // Hope it works... 🤞

The WebMCP Way Reliable

// Direct API - always works const data = await webmcp.query('site://api/about'); // { // name: "Veronica", // type: "AI Intelligence System", // version: "2.4.1", // status: "operational" // } // Clean, typed, reliable ✅

How WebMCP Works

Four stages to AI-native web communication

01

Discover

Site exposes WebMCP endpoint

02

Describe

Schema defines available queries

03

Query

AI requests structured data

04

Respond

Clean JSON returned

Two API Types

Declarative for simple forms, imperative for complex flows

Declarative APIs

For simple forms

Static queries with predictable responses. Perfect for product catalogs, pricing pages, and documentation indices.

GET /webmcp/products
→ [{id, name, price, ...}]

Imperative APIs

For complex flows

Multi-step interactions with JavaScript-heavy applications. Handles auth, workflows, and dynamic content.

POST /webmcp/flow/checkout
→ {sessionId, steps, state}

Try WebMCP

Interactive terminal demo

webmcp — terminal

$ webmcp discover veronica.ee

✓ WebMCP v1.0 endpoint found

✓ Schema: 24 query types available

✓ Auth: none required (public)

$ webmcp query veronica.ee://about

{

"name": "Veronica",

"type": "AI Intelligence System",

"capabilities": ["research", "analysis", "strategy"],

"version": "2.4.1",

"status": "operational"

}

$ webmcp query veronica.ee://docs/endpoints

[

{ "method": "GET", "path": "/v1/research", "auth": false },

{ "method": "POST", "path": "/v1/query", "auth": true },

{ "method": "GET", "path": "/v1/status", "auth": false }

]

Why WebMCP?

Benefits for AI agents and website owners

89% Fewer Tokens Structured data vs parsing verbose HTML

97.9% Success Rate Reliable schemas vs brittle selectors

53% Lower Cost Fewer API calls, less processing

Type-Safe Self-describing schemas prevent errors

Real-Time Updates Change notifications for agents

Universal Protocol One standard, millions of services

Roadmap

The path to AI-native web

Q1 2025

Protocol Spec v1.0

Core specification and schema definitions

Done

Q2 2025

Veronica Native Support

First-class WebMCP integration in Veronica

In Progress

Q3 2025

WebMCP Registry

Public registry of WebMCP-enabled sites

Planned

Q4 2025

Platform Adoption

Major CMS platforms add WebMCP support

Planned

Ready to Build the AI-Native Web?

Join the next evolution of the web — built for AI, not just humans.

Read the Docs Contact Us