Blog

How I think about
building AI systems

Practical writing on automation architecture, production engineering, and the decisions that separate systems that hold up from ones that don't.

Why most AI automations fail in production
The demos work. The production systems don't. Here is what actually goes wrong and how to design around it from the start, not after the first failure.
AI automation vs hiring a VA: how to think about the choice
A founder asked me how AI automation is different from just hiring a virtual assistant. The honest answer is more nuanced than most people expect, and the decision matters more than the cost comparison.
A system without a maintenance plan is not a finished system
Most automation projects are considered done at delivery. They are not. Here is what a real maintenance plan looks like and why it belongs in the architecture, not the afterthought.