AI BUILDER · TRIATHLETE · FATHER · BUILT BEEKEEPER TO 7M USERS
I build AI systems. For fun, for business, and to run my own life.
Twelve agents, two data lakes, one rule: no slides — only things that
survive contact with reality. Built Beekeeper to 7M frontline users and an exit.
Now back building with AI.
FIND THE BIGGEST PROBLEM. CREATE FROM ZERO. MASTER IT. TAKE IT TO THE LIMIT.
System statusAugust 2026
8LIVE2BUILDING1DISCOVERY
MORNING CHAINRUNS DAILY · 07:30
What am I building
Nothing in that category yet.
Building notes
Every system on that shelf has a story that didn't fit on the card — what I
tried first, what broke, and the thing I only understood afterwards. I'm writing them as I
go, one at a time.
How I replaced a $750/year subscription in 3 hours
MONEYLAKE
SOON
Scoring a road for how likely it is to hurt you
DERISK MY RIDE
SOON
What changed when AI started planning our family's food
HOME COOKER
SOON
More detail lands here as I build. Stay tuned.
How I build
No slides.
If it hasn't run in production against my real life, it's an opinion.
Never trust the output.
AI is amazing. But you can't trust that what comes back is exactly what you asked for.
No testing, no releasing.
Automate the boring, keep the judgment.
The agents decide what I look at. I decide what I do. That line never moves.
Own the data.
Every system here writes to something I control. Subscriptions end. Companies pivot.
Fourteen years of my own data doesn't.
Story
I'm a chemical engineer who ended up building and leading a software company. Fourteen years of Beekeeper, from
an idea to 7M frontline users and an exit. That's a long race — and a long race
teaches you fast what actually works and what only sounded good in the meeting.
Now I'm on the other side of it. No company to run — on purpose. A bench full of systems I
built for myself, a training plan I engineer like a product, and the time to work out what
sustains this rather than what replaces it. This page is the honest log —
including the parts that broke.
Or a crazy vision you've always wanted to make real? Those are my favourite. If you're sitting on something everyone calls unsolvable and
you'd rather build than buy, let's talk. I take on two or three at a time, hands on the
keyboard, building next to your team — not writing a deck about what they should do.
Working sessions, not a slide deck about AI. We bring your real problem into the room,
pull it apart, and rebuild it live. Your people walk out having built something — not
having watched me build something.
Scaling to 7M frontline users. What the exit didn't change. And what happens when you
build AI systems as an entrepreneur and operator. English, German or Spanish —
whichever the room speaks.