I built Danubia, a web scraping and content extraction API, hosted in the EU. Why: I used to build crawler infrastructure for a search engine company. I wanted to apply my knowledge to building a clean text extraction API, in part because was it fun to do (why else do we program things, if not for fun?) and in part because I felt I could build something good: a scraping API, with genuinely good content extraction, hosted in the EU, and no subscription needed to use it. It does this: You give it a URL, and it gives you back clean markdown (or HTML, or plain text), no ads, no boilerplate. You can do that on-demand, or you can do that in batch. If you do it in batch, you save credits. The whole thing is backed by a shared crawling layer, which is capable of enforcing politeness (concurrent downloads, robots.txt filtering, crawl delay and retries) at scale. The content extraction layer is a library which I built from scratch. I used Trafilatura as a starting point but rewrote it so I could own the extraction process and improve it continuously. I run a benchmark of my extraction lib against Trafilatura, Readability, Turndown, and Firecrawl and try to do match if not beat them. The tech stack is Typescript + Effect. For now, I'm offering free credits to anyone who would like to try it. In the future, I'd like to offer paid credits top-up. No subscription. I am in a validation phase, so I'm not charging for credits. Anyone who signs up get 500 credits immediately, no credit card needed. I only wish to collect some feedback. If you build something with Danubia, message me and I will give you more credits. I'm also really curious to know what you're building! I also a clear roadmap for the next steps: I don't want to stop at clean text extraction, I want structured data extraction that turns URLs into clean datasets at scale.

Danubia

Turn any web page into clean markdown. Hosted in the EU

Turn any web page into clean markdown. Hosted in the EU