Open-source framework for building AI agents to navigate websites
Each tool is evaluated against our methodology using public docs, vendor demos, and hands-on testing.
What is LaVague?
LaVague is an open-source (Apache-2.0) framework for building AI agents that navigate websites autonomously and perform actions. It combines a World Model, which turns an objective and the current page state into instructions, with an Action Engine that carries them out through Selenium or Playwright. Documented use cases include converting Gherkin specs into automated browser tests, filling out forms and navigating across SaaS apps, and extracting data from web applications.
Our verdict
The World Model and Action Engine split gives LaVague a clear way to turn a plain-language objective into concrete browser actions, and support for local or on-VPC models (including Llama 3) or Azure OpenAI suits privacy-sensitive deployments. No funding or customer case studies are disclosed, so teams still need to run their own tests against a target site rather than lean on outside validation.
Categories:
Agentic automation tools give an LLM control of a browser to navigate, click, and fill forms toward a goal, rather than following a fixed script. They are newer and less predictable than selector-based scraping, but they handle sites and workflows that are hard to hardcode.
Open source AI browser automation SDK from Browserbase
Freemium·Jul 2026·Agentic Automation
Open-source browser agent framework plus a paid cloud API
Freemium·Jul 2026·Agentic Automation
Open source browser automation agent using LLMs and vision
Freemium·Jul 2026·Agentic Automation
How LaVague compares
Browser UseBrowser Use is a comparable open-source Python library that lets an LLM drive a browser directly, without LaVague's separate World Model and Action Engine split.
StagehandStagehand is another open-source framework for AI-driven browser automation built on Playwright, aimed at the same natural-language-to-browser-action problem.
SkyvernSkyvern addresses the same browser-automation-via-LLM use cases but is built around vision-based UI understanding rather than LaVague's instruction-execution engine.
