Howdy from Madrid

Hi everyone, I’m Rubén, based in Madrid. I started out in translation and machine translation, spent a few years as a data scientist at PayPal, and now work as a data architect. I also teach programming to people with humanities backgrounds.

Right now I’m building conversational analytics with ClickHouse, dbt, and Strands Agents. The question I keep coming back to is how to let people ask questions in their own words while making sure the answers use metrics everyone agrees on.

I’d love to swap notes with anyone building AI workflows for real teams. What worked in your demo but needed a rethink once people started using it?