>> Your AI Reads Customer Data — A Proxy Solved My GDPR Problem
Four approaches to GDPR-compliant AI. Three require rebuilding your pipeline. A pseudonymization proxy keeps frontier model quality and ships in days.
The decisions you won't find in the docs. Trade-offs, gotchas, and the stuff that's obvious in hindsight.
German PaaS options exist, but none match Heroku-level simplicity. k3s on Hetzner gave me scalable infrastructure with git-push deploys and full observability — without leaving the country.
FluidAudio, MLX, ScreenCaptureKit, and zero network calls. How I shipped a macOS transcription app that runs entirely on your machine.
How dotenvx eliminates custom CI code for managing Supabase credentials across environments — and why encrypted secrets in git might be fine.
Native Messaging options for connecting browser extensions to desktop apps — and why I picked Unix sockets over HTTP.