Find and build AI solutions for your website and web application.
Discover task-based APIs built in the browser that perform AI tasks with local models.
Create WebMCP tools so sites can support reliable action by AI agents.
Improve compute and model performance with Wasm, WebGPU, and WebNN.
COURSE

Evaluations (or evals) are tests for your AI features. When you create a robust evals pipeline, your LLM-based features will work well for your users. Then, you can ship your features with confidence.

In this course you'll learn how about the LLM-as-a-judge pattern to measure subjective quality, create a judge with minimal setup, and discover advanced techniques to develop a custom judge that thinks like top-domain experts. You'll run evals at build time and in production so you can catch regressions early.

Start learning

Perform inference locally, in the browser

Build features powered by AI on the web, including foundation models and expert models, built into the browser
Discover built-in AI
Learn the requirements to start building features and applications with built-in AI.
Review the APIs and their availability status to use with built-in AI in Chrome.
Use the built-in AI task API polyfill to build AI features, without worrying about browser support.
Learn the best practices for building with built-in AI in Chrome.
Learn how to cache models in the browser for faster inference.
Before any of the APIs can be used, the underlying model must be downloaded. Always alert users about the time required to perform these downloads.

Reliable agent calls with WebMCP

WebMCP aims to provide a standard way to expose structured tools, so that AI agents can perform actions on your site with increased speed, reliability, and precision.
Get started
Step 1

Your site declares the actions an agent can take, as WebMCP tools. Each tool has a name, description, and JSON schema.

// Imperative API
document.modelContext.registerTool({
  name: "bookSlot",
  title: "Book a consultation slot",
  description: "Reserve a 30-minute consultation.",
  inputSchema: {
    type: "object",
    properties: {
      date:  { type: "string", format: "date" },
      time:  { type: "string" },
      name:  { type: "string" },
      email: { type: "string", format: "email" }
    },
    required: ["date", "time", "name", "email"]
  },
  annotations: { readOnlyHint: false },
  execute: async (input, client) => {
    return await api.book(input);
  }
});
Step 2

The browser collects and presents these tools to the user's WebMCP-aware agent, alongside the page's URL, title, and origin permission scope.

Your site speaks to the browser, which translates for the agent.
Step 3

The agent sees a contract, supplies structured arguments, and your code does the rest. The user stays in the loop for permission and confirmation. No more guessing.

// the agent's view
{
  "tool": "bookSlot",
  "arguments": {
    "date": "2026-07-31",
    "time": "14:00",
    "name": "Jerry Jones",
    "email": "jerry.jones@example.com"
  }
}
When to use WebMCP Build effective tools

Improve compute and model performance

Learn more

Learn about improvements to WebAssembly for tasks commonly used in AI inference.

Read the doc Chrome Status

Discover WebGPU in Chrome and get first-class support for general computations on the GPU.

Read the overview Chrome Status

Use the WebNN API to perform machine learning operations in the browser.

Read the spec Chrome status
COURSE

Whether prompted by company mandates or enthusiasm, AI for AI's sake isn't useful. The best features come from a user need and are measured by the value they deliver.

In this course, you learn how to think about building web applications with an AI system that is responsible, usable, and valuable, so that your application delivers on the technology's promise.

Start learning

Coding assistance

Download skills for your AI coding agent to guide its generation towards modern and performant experiences.
Chat with Gemini in DevTools. Conversations you start from this panel have context about technical details of the page you're inspecting.
Use the Console to view and debug the results of on-device AI.

Meet the Team

The Chrome Web Developer Relations team and built-in AI PM are available to chat about the web and AI from built-in to WebMCP. We have folks in Asia, Europe, and the America

Meet the team

AI and Gemini Nano news

Introducing web AI

web.dev
Understand the basics and definitions of the various emerging technologies, often referred to as AI.
Read the doc
web.dev
As web practitioners, it's critical that we build new technology thoughtfully and responsibly.
Read the doc
web.dev
Discover techniques to use client-side AI, which offers benefits such as low latency, reduced server-side costs, increased user privacy, and more.
Read the doc