Browser Use is a Python library that gives AI agents control of a real browser. The agent receives a task in natural language, inspects the page, decides on clicks and types into fields until it finishes.
What it is for
- Automating tasks on sites that offer no API
- Extracting data from dynamic pages that require interaction
- Filling forms, navigating portals and collecting recurring information
- Testing web flows with an agent that adapts to layout changes
How it works
- The agent combines a language model with a tree of page elements
- Available actions include clicking, typing, scrolling, extracting content and running JavaScript
- It works with major model providers, including local options
- The cloud version offers managed browsers and an API to launch tasks
- It supports multiple tabs, persistent sessions and run recording
Availability
- Open-source library under the MIT licence
- Commercial cloud service billed per browser hour and model usage
Points to note
- Sites with blocks, captchas and verification lower the success rate
- Acting on real accounts calls for dedicated credentials and scope limits
- Long tasks get expensive in tokens; measure cost per run