Hey Product Hunt, I've been connecting LLMs to browsers for a while now, and it's always the same story. You spin up Playwright or Puppeteer, the agent gets back a wall of raw HTML, burns most of its context window on markup noise, and then you're guessing wait times, building CSS selectors that break next week, and dismissing cookie banners manually. The existing browser APIs (Browserless, Browserbase, etc.) solve the infrastructure part, but you still get raw HTML back. The LLM still has to parse it. So I built Browserbeam. It's a REST API that returns what LLMs actually need: * Markdown instead of HTML * Short element refs (e1, e2) instead of CSS selectors * A stability signal so you know when the page is ready * A diff after every action showing what changed * Cookie banners and popups dismissed automatically * Declarative extraction: describe the shape, get JSON You send JSON, you get structured JSON back. One POST request replaces ~25 lines of Puppeteer code. There are official SDKs for Python, TypeScript, and Ruby. Also an MCP server if you use Cursor or Claude Desktop. Free trial with 1 hour of runtime, no credit card required. Would love to hear what you think, especially about the API design and whether the response format makes sense for your use case. https://browserbeam.com

Browserbeam

The browser API built for AI agents

The browser API built for AI agents