I’m a product manager who builds the AI tools I write about. Everyone’s working with the same handful of models now; the difference is what you build around them. That’s the whole subject here — the context, the guardrails, the systems, the unglamorous part that turns a model into something people actually use.
I write for people who want to start building but can’t find the door. Not the folks already shipping agents at work, but the ones who suspect this is within reach and just need someone to show the path, without selling them a course, a framework, or a five-step funnel.
Everything here is free, and the resources are open source. If a piece helps and you want to support the work, pledges are open. They fund the time, not a paywall. The content stays free either way.
New here? Start with these:
The memory series — how AI memory actually works, why it isn’t yours, and why the obvious fix still isn’t safe. Three parts: Your AI’s Memory Isn’t Yours, Anyone Can Write to Your AI’s Memory, and Cites Its Sources Now.
Your AI Tool is a Service — why the model is a commodity and the product is everything around it. The argument this whole thing runs on.
Most AI therapy tools describe the modality. They don’t run it —the other thing I build: AI for mental health, and the gap between naming a method and running it.
Musician, recovering audio engineer, congenitally skeptical of anything that demos too well.
