As of today, February 20, 2026, this is how I use AI/LLM in my workflow.
I like to get better at writing as it sharpens my thinking. So I don’t have AI write the content on this website for me (I know, the irony), so my understanding of AI improves faster.
I do use AI to critique structure and clarity, but I always own the words.
The point I make to let you know I do the writing on this website myself, is me struggling with my integrity. And why it matters to me in a world flooded with AI slop.
Tooling
There is my OpenClaw agent (Henk) running that I interact with while working on this website. And I ask a variety of LLMs questions via the respective apps and via tools like Codex and Claude Code.
When programming, I mostly vibe code things, with minimal inspection. I do not want to learn how to code, I want to build and ship things. I do understand I give up control over architecture and security. I also give up deep understanding. The trade off for making things more interesting and failing quickly.