Quickstart

Build an AI-powered data extraction agent that pulls product data from e-commerce sites and stores it in MongoDB with automatic schema validation and data analysis capabilities.

What You'll Build

An AI-powered data extraction agent that pulls product data from e-commerce sites and stores it in MongoDB with automatic schema validation and data analysis.

Before you start

Ensure you have these requirements ready:

Node.js 22.18+

Required runtime environment

MongoDB

Local install or MongoDB Atlas

API Access

Browserbase API key

Step 1: Project setup

Clone and install

# Clone the integration template
npx degit browserbase/integrations/examples/integrations/mongodb/typescript browserbase-mongodb
cd browserbase-mongodb

# Install all dependencies
npm install

# Install browser binaries
npx playwright install

What gets installed?

Core packages:

• @browserbasehq/stagehand - AI-powered data extraction
• mongodb - MongoDB driver for data storage
• zod - Schema validation for type safety

Utilities:

• chalk & boxen - Terminal styling and output formatting
• playwright - Browser automation engine

Step 2: Start MongoDB

Required: MongoDB must be running on your system before proceeding. Start it with mongod if installed locally, or ensure your MongoDB Atlas connection is ready.

If using local MongoDB, ensure it’s running:

# Start MongoDB (if installed locally)
mongod

MongoDB Atlas users can skip this step as the database is already hosted in the cloud.

Step 3: Configuration

Environment variables

Create your .env file with the required configuration:

Local MongoDB

# Browserbase Configuration (Recommended)
BROWSERBASE_API_KEY=your_browserbase_api_key

# MongoDB Configuration
MONGO_URI=mongodb://localhost:27017
DB_NAME=scraper_db

Step 4: Configure Stagehand

The integration uses Browserbase cloud browsers:

Node.js stagehand.ts

import { browserbase, Stagehand } from "@browserbasehq/stagehand";

const browser = await browserbase.launch({
  apiKey: process.env.BROWSERBASE_API_KEY,
  browserSettings: {
    blockAds: true,
    viewport: { width: 1024, height: 768 },
  },
});

const stagehand = await Stagehand.create({
  browser,
  logging: {
    level: "info",
    format: "pretty",
  },
});

Step 5: Run your first extraction

What happens when you run the agent:

1. Connects to MongoDB and creates necessary collections
2. Navigates to Amazon laptop category
3. Extracts product listings with AI-powered data extraction
4. Extracts detailed information for the first 3 products
5. Stores all data in MongoDB with schema validation
6. Runs analysis queries and displays results

Execute the extraction

npm start

Customization options

Extend data schema

Add custom fields to capture more product information:

Node.js

const ProductSchema = z.object({
  // Existing fields...
  title: z.string(),
  price: z.string().optional(),
  rating: z.string().optional(),

  // Add your custom fields
  brand: z.string().optional(),
  availability: z.string().optional(),
  shippingInfo: z.string().optional(),
  specifications: z.array(z.string()).optional(),
  customerReviews: z.number().optional(),
});

Custom extraction instructions

Modify the AI extraction to capture specific data:

Node.js

const data = await stagehand.extract(
  `Extract comprehensive product information including:
    - Brand and model details
    - Detailed specifications
    - Availability and shipping information
    - Customer ratings and review counts`,
  ProductSchema,
);

Expected output

The extraction stores product records in the configured MongoDB database and prints analysis results. Inspect the stored documents to confirm that the fields match your schema; product values and availability vary by run.

Cleanup

Confirm the browser session ended, and release it from the Sessions dashboard if necessary. Remove only the tutorial records from scraper_db when you no longer need them. If you started a local mongod process for this tutorial, stop that process with Ctrl+C; leave shared databases running.

---

Need help? Join the Stagehand Slack community for support and to share your projects!

Related guides

Scale your extraction

Learn how to scale your data extraction across multiple sites and handle larger datasets.

Deploy to production

Deploy your extraction pipeline to production with Browserbase.