I build software for problems I know firsthand.
A finance system I use every day. A booking system being developed for a small rental business. Built and maintained with AI agents.
Selected projects

Finance
In daily use
What's left this month, cash ahead, and recurring bills that flag their own price and name changes, shaped around my own needs.
Its central rule: never overstate available cash or present stale data as current.
Shown with demo data. Captured .

Journey
In owner testing
Booking and owner workflows for a camper rental business, built around how the business works.
The owner has a demo to try. It isn't taking real bookings yet.
Captured .
How I build and learn
From a need to working software
Agents write the code. I work through the requirements, question unfamiliar decisions and try the resulting software. Previews, independent reviews, automated tests and recovery checks help me assess changes.
When something breaks or doesn't fit, I use that experience to refine the system and my understanding of it.
Learning through the work
Working this way lets me pursue custom software that previously would have been impractical for me (or possibly anyone else for that matter) to build.
I'm learning how software fits together, which decisions to delegate and what evidence makes the result trustworthy. When I don't understand something, I question the agent until I understand enough to move forward.