My problem was that I found these companies hard to discover unless you already know where to look or have the right connections.
So I started mapping them manually. Belgium first, then the Netherlands, then it kept growing.
Today startupmap.one has 2,300+ curated European startups with live hiring data, funding stages and locations. Built first with Lovable, but recently migrated to Claude Code. I have no CS experience, and have never written a line of code.
Some stats after the first 90 days: - 18,000+ monthly visitors - 6+ minute average session - 683 applications started on the platform - 1 in 3 visitors clicks through to a startup's careers page
Traffic came mostly from Reddit posts in city subreddits & LinkedIn posts. Word of mouth picked up after that. Now starting to invest more time in SEO/GEO.
What I actually had to learn: Even with zero coding background, I had to genuinely understand how Supabase, Neon, and APIs work. Not to write the code, but to make good product decisions. Claude was an incredible tutor for this. It's not just an LLM, it's a designer, a tutor, a marketing genius, a senior dev, it's my co-founder at this point.
The hardest problem: The performance. 2,000 logo PNGs being loaded made the website very unsatisfying to navigate. I spent multiple sessions trying to get it to cluster properly to improve performance. Claude kept suggesting approaches that didn't fully work. Eventually I took a step back, thought of a way simpler solution (one it hadn't suggested). That worked. My takeaway: AI can execute really well, but thinking critically about the problem yourself before handing it over is often more efficient.
The scariest moment in the earlier days: a dev friend (luckily) pointed out I hadn't enabled RLS on my database, meaning anyone could have wiped everything. oops.
Happy to answer anything about building a directory product with zero technical background.
Also interesting would be that this is map-based, so physical location but I think you will find a lot of startups with unclear physical locations.
They're using "AI" \i :)