Writing
Thoughts on AI, software architecture, and building products.
Lab reports are databases sent as PDFs
External lab test results often arrive as PDFs in different formats. AI-assisted parsing can turn them into structured data for compliance, quality management, and analysis.
Laborberichte sind Datenbanken, die als PDF verschickt werden
Externe Laborberichte kommen oft als PDF in unterschiedlichen Formaten. Mit KI-gestütztem Parsing lassen sich daraus strukturierte Prüfdaten für Compliance, Qualitätssicherung und Analysen gewinnen.
EDI in 2026: Standards, Protocols, and the Full Process Chain Explained
A technical deep dive into EDI for IT managers and CTOs in mid-market companies — standards, transmission protocols, partner onboarding timelines, and where AI fits in realistically.
EDI Mapping Tools in 2026: Who Actually Lets You Build Your Own Mappings?
An insider analysis of EDI mapping tool vendors — who offers real self-service mapping, who is a managed service in disguise, and what the Lobster PE acquisition means for 2,000+ customers looking for alternatives.
The Real Cost of EDI: What the Pricing Page Doesn't Tell You
EDI vendors quote per-message fees and platform subscriptions. The actual cost is 2-3x higher when you account for mapping, testing, maintenance, and personnel. A structured breakdown for CTOs and CFOs evaluating EDI spend.
SAP System Analysis with AI Agents in 2026
SAP's ADT has a full REST API. Combined with OData and an LLM, an AI agent can scan custom code, run ATC checks, and map your system in hours — not weeks. A technical deep-look for 2026.
Wie KI Enterprise-Reporting verändert — vom Flaschenhals zum Self-Service
KI verändert nicht was wir reporten — sondern wer Reports erstellen kann. Warum der eigentliche Flaschenhals im Reporting der Prozess ist, nicht die Technik. Und wie Self-Service-Reporting mit KI in der Praxis funktioniert.
Claude auf AWS vs. Azure vs. Google Cloud: DSGVO-Datenresidenz im Vergleich (2026)
Ein praxisnaher Vergleich von Claude AI auf AWS Bedrock, Azure AI Foundry und Google Vertex AI — mit DPA-Analyse, Data-Residency-Optionen und einer pragmatischen Compliance-Strategie.
Claude on AWS vs. Azure vs. Google Cloud: GDPR Data Residency Compared (2026)
A hands-on comparison of hosting Claude AI on AWS Bedrock, Azure AI Foundry, and Google Vertex AI — with DPA analysis, data residency options, and a pragmatic compliance strategy.
AI Agent Mac Mini: How I Run an Always-On AI Infrastructure for €3/Month
A practical guide to running an always-on AI agent on a Mac Mini M4. Real setup with OpenClaw, costs vs cloud, and what 20 autonomous cron jobs actually do.
Why Cheaper AI Makes Product Thinking More Important, Not Less
AI inference costs are collapsing. That's great for builders — and terrible for anyone whose only moat is an API call. Here's why product thinking is the real differentiator.
Day 5 With My AI Employee: What Happens When You Give an Agent Full Access
I gave an AI agent full access to my systems from day one. Five days in, here's what's actually working, what's not, and what I'm learning.
RAG for RFP Responses: My Architecture Plan for BidScribe
I'm building an AI-powered RFP tool that needs production-grade RAG. Here's my architecture plan — what I'm prototyping, what I've learned from research, and what I think will actually work.
KI mit SAP verbinden: 3 Architektur-Optionen, die tatsächlich funktionieren
ChatGPT und Claude mit SAP-Daten verbinden klingt einfach — ist es aber nicht. Drei Architektur-Optionen im Vergleich: Self-Hosted, SAP BTP und Cloud-Hosted.
Building BidScribe: Why I'm Tackling RFP Responses with AI
I'm building an AI-powered RFP tool as a side project. Here's the idea, the planned tech stack, and why I think this problem is worth solving.