Notes on design
& building.
Practical writing on design careers, AI products, and what happens when a designer ships apps alone. No fluff, no process theater.
-
July 2026 · Careers in design · 9 min read
How to become a product designer in 2026 — advice from 14 years in
My degree is in tourism. Here's the actual path into product design — the skills that matter now, a portfolio with zero experience, the first job, and an honest take on the AI question.
Read the guide → -
July 2026 · Building in public · 8 min read
I shipped two iOS apps solo — what AI changed about being a designer
After 14 years designing products other people built, I took Numi and Tysha from idea to the App Store alone. What AI tools actually changed — and what they didn't touch at all.
Read the story → -
July 2026 · AI & design · 7 min read
Designing AI products: 6 lessons from building an AI nutrition app
AI products are probabilistic — sometimes wrong, always uncertain, never instant. Six reusable UX lessons on uncertainty, latency, correction loops and failure states, from a shipped product.
Read the lessons → -
August 2026 · Design practice · 7 min read
AI product design: designing for the cost of being wrong
AI product design is ordinary product design plus one new variable: the feature can be confidently wrong and you cannot tell which time. I name the wrongness budget, separate draft surfaces from commit surfaces, and go through the decisions I actually changed after shipping.
Read the piece → -
September 2026 · Design & ML · 8 min read
UX for machine learning: the interface is a measuring device
Machine learning products have three time horizons, and most design writing only covers the first one. Here is what I learned shipping ML features about correction paths, checkability bias, and the fields that should never ask for confirmation.
-
September 2026 · AI interfaces · 11 min read
AI UX patterns that survive contact with a real model
A working list of the AI UX patterns I actually use, with the reasoning behind each one and the versions that failed first. Drawn from building and shipping Numi, a vision-based food tracker, on my own.