Zum Hauptinhalt springen

Blog

Gedanken zu AI, Software-Architektur und dem Bauen von Dingen, die funktionieren.

9 Beiträge · seit 2025

Der Mittelstand-Pfad: Warum DACH-Unternehmen eine andere KI-Coding-Strategie brauchen als das Silicon Valley
de

Der Mittelstand-Pfad: Warum DACH-Unternehmen eine andere KI-Coding-Strategie brauchen als das Silicon Valley

Silicon-Valley-Playbooks für KI-Coding-Tools passen nicht zum Mittelstand. Fünf Punkte für 15-Jahre-Monolithen, BaFin und Betriebsrat.

Weiterlesen →
Der Betriebsrat und Ihr KI-Coding-Rollout: Das Gespräch vorbereiten, bevor es eskaliert
de

Der Betriebsrat und Ihr KI-Coding-Rollout: Das Gespräch vorbereiten, bevor es eskaliert

Claude Code, Cursor, Windsurf: der Rollout wird blockiert, beide Seiten sind frustriert, niemand versteht recht, warum. Ein Briefing für Engineering Leads, die den nächsten BR-Termin nicht verlieren wollen.

Weiterlesen →
The Case for a Programmable Desktop
en

The Case for a Programmable Desktop

LLMs doubled the number of windows on my screen. The old drag-and-snap managers could not keep up. A tiling window manager, a remapped CapsLock, and a bash script that hunts down music-playing tabs turned out to be the fix.

Read more →
Who Owns the Means of Computation?
en

Who Owns the Means of Computation?

Tokens are the new unit of work. The deeper question: who's the landlord? A look at the three layers of control in the token economy, and the one wrinkle that complicates everything.

Read more →
Zen and the Art of AI Maintenance
en

Zen and the Art of AI Maintenance

Robert Pirsig wrote about alienation from technology. Fifty years later, the same pattern plays out in enterprise AI adoption: benefits without understanding, capability without diagnostic literacy.

Read more →
The Magic Words That Make AI Code Better
en

The Magic Words That Make AI Code Better

Deep domain knowledge, expressed through precise terminology, dramatically improves AI output. Here are the magic spells that actually work.

Read more →
The Week Everyone Decided AI Agents Were Dead - (Except They Weren't)
en

The Week Everyone Decided AI Agents Were Dead - (Except They Weren't)

Everyone read the headlines about AI failure wrong. The 80% that fail and the 20% that thrive use the same technology. The difference? Boring architecture over magic thinking.

Read more →
Good Taste in Software Engineering: Tests, Not Vibes
en

Good Taste in Software Engineering: Tests, Not Vibes

Taste isn't mysterious. It's disciplined judgment about tradeoffs, with anchors, scorecards, and practices you can learn today.

Read more →
As we may prompt
en

As we may prompt

Or: when AI Commoditizes Execution, Attention Becomes Your Only Moat

Read more →