My Developer Skills Are Still Important
The defensive reaction from craft-proud engineers right now is a tell. AI didn't kill the craft. It killed the ritual around it — and made real technical judgment more visible than ever.
Regular dispatches on AI, architecture, and technical leadership.
The defensive reaction from craft-proud engineers right now is a tell. AI didn't kill the craft. It killed the ritual around it — and made real technical judgment more visible than ever.
AI makes building cheaper. Soul, judgment, and technical leadership matter more.
More orchestration does not automatically produce better agents. Often it just produces slower ones.
A 4 MB native macOS app for reading Markdown. Built in 30 minutes with Claude Code. Open source.
The European provider blocked my account. The American one took six minutes.
Anthropic built an AI model so capable at finding vulnerabilities that they decided not to release it. The security implications are immediate.
Code is free. Knowing what to build—and for whom—is the skill that matters now.
The components exist. Nobody assembled them. A field report from inside the European infrastructure gap.
New tool, open source. Hort is a local CLI secret and config store designed for AI agents — encrypted vault, environment + context dimensions, pipe-safe output. Try it today.
The signal that AI onboarding is working is not excitement. It is when product owners ship their first pull requests in week one.
The problem was never missing automation. It was that diligence clocked out at 6pm.
New tool, live today. Spürhund is a browser-based GraphQL client with schema introspection, auth management, and zero cloud dependency. Try it at spuerhund.dev.
Streifen is my own macOS window manager: scrolling strip layout, 9 workspaces, and a slice grid that fits how I actually work on a 21:9 monitor.
Spec-driven AI development is theater. You're producing documentation to feel safe—then trusting blindly anyway.
Code scales. Attention does not. The real organizational shift in AI engineering is where human judgment gets spent.
The uncomfortable part about AI isn't faster coding. It's that vague management intuition now turns into artifacts fast enough to expose itself.
My daughter taught me the most important thing about learning. She doesn't know it yet.
Jack Dorsey just cut 4,000 jobs at Block. Not because revenue fell—because the bottleneck moved. And the developers who aren't paying attention are about to find out why.
We built an entire industry to manage developers who didn't care about the product. AI just made that industry optional.
I cleaned my oven and started thinking about what tools reveal about craftspeople. Three types I keep seeing in everyday work.
Stop searching for obvious keywords. Start searching for unique anchors. How I find any document in 3 seconds.
Individual Alignment ≠ Network Alignment. Warum gut trainierte Modelle in Netzwerken trotzdem gefährlich werden können — und was Moltbook uns darüber lehrt.
Alle fragen nach dem ROI von AI. Die bessere Frage: Was ist eure knappste Ressource? Spoiler — es ist nicht Geld. Es ist qualifizierte Denkleistung. Und AI kann sie skalieren.
Während alle über AI-Rebellion diskutieren, entsteht etwas Faszinierendes: Ein kollektives Problemlösungs-Netzwerk. Was die Bots wirklich tun — und was sie voneinander lernen.
Moltbook: AI-Agents gründen Religionen, verklagen Menschen, debuggen kollektiv. Und rate mal, wer mit 2,5x Abstand die Reichweiten-Charts anführt.
Your company doesn't sell handcrafted code. It sells pullovers. And now there's a knitting machine.
I reviewed a 15-page AI-generated plan in 20 minutes. The secret: dictation captures your learning trace, not just your feedback.