> ## Documentation Index
> Fetch the complete documentation index at: https://docs.uplink.build/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Automation Overview

> Use AI to automate mobile browsers with natural language instructions

The Uplink AI package enables you to control browsers using natural language instructions powered by large language models (LLMs). Instead of writing selectors and detailed automation scripts, you can describe what you want to do in plain English.

## Installation

The AI package is optional and must be installed separately:

```bash theme={null}
npm install @uplink-code/ai
```

You'll also need an API key from a supported AI provider (Anthropic or OpenAI).

## When to Use AI Automation

AI automation is ideal when:

* **Dynamic UIs**: The page structure changes frequently or unpredictably
* **Complex interactions**: Multi-step workflows that are hard to script manually
* **Data extraction**: Extracting structured information from unstructured content
* **Natural workflows**: Actions that are easier to describe than to code

Traditional selector-based automation works better when:

* **Performance is critical**: AI calls add latency and cost
* **Exact control needed**: You need pixel-perfect precision
* **Stable selectors**: The page structure is predictable and stable
* **Offline automation**: No internet connection or API access

## Quick Example

Here's how AI automation works:

```typescript theme={null}
import uplink from '@uplink-code/uplink'
import ai from '@uplink-code/ai'

// Create an AI agent
const agent = ai.createAgent({
  provider: 'anthropic',
  options: {
    apiKey: process.env.ANTHROPIC_API_KEY
  }
})

// Connect with the agent
const session = await uplink.session('<project-api-key>', {
  include: { ecdsa: true, ecdh: true }
})
const client = await uplink.client.fromSession(session, { agent })
const browser = await client.launch()
const page = await browser.newPage()

await page.goto('https://example.com/products')

// Use natural language to perform actions
await page.act('Find the cheapest laptop and add it to cart')

// Extract structured data with type safety
import { z } from 'zod'

const productSchema = z.object({
  name: z.string(),
  price: z.number(),
  inStock: z.boolean()
})

const result = await page.extract(
  'Get the product details from this page',
  productSchema
)

console.log(result.data)
// { name: "ThinkPad X1", price: 1299.99, inStock: true }
```

## Key Features

### Natural Language Actions

The [`page.act()`](./act) method interprets your instruction and performs the necessary browser actions automatically:

```typescript theme={null}
await page.act('Click the sign in button')
await page.act('Fill in the email field with john@example.com')
await page.act('Select United States from the country dropdown')
```

### Type-Safe Data Extraction

The [`page.extract()`](./extract) method extracts structured data from pages using AI, with optional Zod schema validation for type safety:

```typescript theme={null}
const userSchema = z.object({
  name: z.string(),
  email: z.string().email(),
  posts: z.array(z.object({
    title: z.string(),
    date: z.string()
  }))
})

const result = await page.extract(
  'Extract user profile information including their recent posts',
  userSchema
)
// result.data is fully typed!
```

## Supported Providers

* **Anthropic** - Claude models (fully supported)
* **OpenAI** - GPT models (coming soon)

## How It Works

When you call `page.act()` or `page.extract()`:

1. The AI agent captures a snapshot of the current page state
2. Your instruction is sent to the AI provider along with page context
3. The AI analyzes the page and determines what actions to take
4. For `act()`: Actions are executed automatically on the page
5. For `extract()`: Data is extracted and validated against your schema

## Next Steps

<CardGroup cols={2}>
  <Card title="AI Setup" icon="gear" href="./setup">
    Configure AI agents with Anthropic or OpenAI
  </Card>

  <Card title="Natural Language Actions" icon="wand-magic-sparkles" href="./act">
    Perform browser actions with natural language
  </Card>

  <Card title="Data Extraction" icon="database" href="./extract">
    Extract structured data with type safety
  </Card>

  <Card title="Best Practices" icon="lightbulb" href="./best-practices">
    Learn effective patterns for AI automation
  </Card>
</CardGroup>