Browser Use

Overview

Browser Use is the leading open-source library for connecting AI agents to web browsers. Instead of building brittle CSS-selector-based scrapers, you describe what you want in natural language, and Browser Use handles the clicks, scrolls, form fills, and data extraction. With 85k+ GitHub stars, it’s the most popular browser automation tool in the AI ecosystem.

How It Works

1. Launch a browser — Browser Use controls a Chromium instance via Playwright
2. Describe the task — Tell the agent what to do in natural language
3. Agent acts — The LLM sees the page, plans actions, and executes them step by step
4. Extract results — Structured data extraction from any website

Use Cases

• Form automation — Fill out applications, registrations, and surveys
• Data extraction — Scrape product listings, reviews, or competitor data
• Web testing — AI-driven end-to-end testing of web applications
• Task automation — Book appointments, manage accounts, process workflows
• Research — Navigate complex multi-page research workflows

Getting Started

from browser_use import Agent
from langchain_openai import ChatOpenAI

agent = Agent(
    task="Go to amazon.com and find the best-rated laptop under $1000",
    llm=ChatOpenAI(model="gpt-4o"),
)
result = await agent.run()
print(result)

Example

Task: "Search for flights from NYC to London on Google Flights for next month"

Agent steps:
1. Navigate to google.com/travel/flights
2. Enter "New York" as origin → "London" as destination
3. Set dates for next month → Click search
4. Extract top 5 cheapest results

Result: [{airline: "Norse Atlantic", price: "$289", duration: "7h 15m"}, ...]

Alternatives

• Stagehand — Browserbase’s AI browser automation framework
• Playwright MCP — Microsoft’s browser automation via MCP
• Selenium — Traditional browser automation (no AI)