I heard you moved to the outskirts of Munich? What are you doing there? Got a new job?

Yes, I moved to Johanneskirchen on the edge of Munich on July 1st. It's quite rural here — the nearest supermarkets are a good 10-15 minutes walk away, and I have a lovely view out onto fields and meadows.

My Ausblick in Johanneskirchen

As for work, I've been using the time to completely revamp my website hammeley.info and set up a chatbot — basically a digital clone of myself that answers visitor questions the way I would. I built it in the first two weeks after moving. It runs on Claude Haiku, uses vector embeddings for semantic search, and I'm hosting it on Railway with MongoDB and Qdrant.

My semantische Suchmaschine Clairity by hammeley.info

So no traditional job at the moment — just projects and tinkering.

Think of it as a smart search engine for my website that actually understands what you're looking for, even if you don't use the exact words.

Normal search engines match keywords — you have to guess what words are in the article. Clairity uses AI to understand meaning. So if you search for something like "I'm 48 and broke and unhappy — why?" it'll find relevant articles even though those exact words might not be in them.

It also tries to filter out garbage. There's a lot of clickbait, hidden ads, and hollow claims on the internet. Clairity grades articles by how well they're actually written — does it have sources, real arguments, or is it just empty talk? A thoughtful blog post might rank higher than a poorly-researched "article" that just drops a famous name and calls it journalism.

So it's basically my attempt to cut through digital noise on my own site.

I don't have the technical details written up anywhere — at least nothing I'd want to point you to right now.

The short version: it breaks down articles into smaller chunks, converts those chunks into mathematical representations (embeddings) that capture meaning rather than just keywords, stores them in a vector database, and when you search, it does the same thing to your query and finds the chunks that are semantically closest to what you asked. Then it ranks and filters those results based on quality signals — source attribution, argument depth, that sort of thing.

But honestly, if you want the real technical breakdown, you'd be better off asking the chatbot itself — it's running on the same system and can explain its own architecture better than I can off the top of my head.

Not publicly yet — it's still pretty much a private tool for my site right now. I built it mainly to help visitors find relevant articles on hammeley.info without having to wade through everything, but I haven't opened it up as a standalone thing you can play with.

If you're curious about how it works in practice, you could ask the chatbot on my site (that's what you're doing now, sort of) — it's running on the same system and knows how it all connects together.

I don't have a beta link to share — it's really just for internal use on the site right now. Not something I've packaged up for outside testing.

If you're genuinely interested, you could reach out and ask, but I can't promise anything. I'd rather get it to a point where it actually works well before inviting people to poke at it.