DesignOps & AI
How the work gets made
Not theory — practice. Short articles on the operating systems behind my projects: the process rituals, the research honesty, and the AI-assisted workflows I use to design and ship. Every article points at the work where it happened.
Demo, not memo
Claude Code turned our Figma designs into a walkable product. Users stopped following the script — and the feedback got real.
Write the system, then the screens
Design systems used to scale a designer across a team. Now they scale judgment to machines — if you write them down.
The app is the HMI
Gardena’s Bluetooth products have no buttons and no screens. When the app is the only interface, configuration time is a hardware spec.
Speak job-site, not catalog
User validation renamed half the Diamond Tool Selector’s interface. Users’ words beat product words — every time.
Invisibility as a KPI
The best registration is the one users don’t notice — value first, scan don’t type.
A designer who ships working software
AI-assisted delivery turned a design studio into a build team — this website is exhibit A.
AI as design material
Shipping my own AI product taught me where models help people — and where they only demo well.
Protect discovery under delivery pressure
Before any screens: an operating rhythm that gives discovery, ideation and iteration room to breathe.
User stories, not feature lists
A roadmap that states what people gain — traceable from demand to story to release.
One service, three audiences
Multi-role experience journeys: the service-design method behind an iF-awarded product.
Label what you don’t know
A placeholder persona, declared honestly, kept a defense project honest.
Conversations as theatre
Roleplay workshops test voice interfaces before a line of code — and catch what flowcharts can’t.
Workshops that start with walls
Co-creation works when experts know the constraints first. Don’t let anyone run into a wall.