I use AI to do real work,
and write down what happens.

Twenty years building digital products, ten of them on productivity software, now spent finding out what AI does for the work you already have. I ship with it every week and publish what I learn, including the parts that fail.

I was a lead engineer on Wunderlist before Microsoft acquired it in 2015, then a Principal Product Manager there. I co-founded informed, a news company backed by HV Capital and 468 Capital. Today I am CTO of Umane, an athletic health club in Berlin where I build the platform and the digital twin behind it, and co-founder of Supercharged, where we run AI workshops for teams.

Most of my week is spent writing code with these tools. I have put several apps and platforms into production that way. Not prompting an assistant for snippets: building the context, the tools and the guardrails a model needs, then running it against a real codebase until it holds up under load.

I keep a harness of my own for this and rebuild parts of it most weeks as the models change. Every release gets tried on something real rather than a benchmark, which is how you find out that the thing everyone is excited about breaks on your data. That is where the writing comes from.

What Actually Works

Every week I take one real task, hand as much of it as I can to AI, and write down what happened, with the prompts, the cost, and where it broke.

Berlin used to fill the gap with savings. Now it borrows.: Berlin plans to spend €45 billion in 2026. Where the money comes from, where it goes, and why the bill for interest rises faster than almost anything else, told in 4,500 dots.
The lots I run past: Berlin is arguing about buying back apartments. I wanted to know how many new ones would fit on the empty land I jog past, so I checked every parcel in the city in an afternoon.
Every correction becomes a rule: Eight months with an AI chief of staff that lives in my notes, and the habit that made it mine.
It digs, I sequence: How saved reels, my Rekordbox library and seven DJ personas became a setup that finds and critiques music for my sets. It may not order the set.
Pia drafts, the team sends: A document search bot for one practice became an inbox assistant that drafts every reply and checks insurers' cuts with sources. It has no send button.
The old MacBook that never sleeps: My AI system runs on an old MacBook at home, and most of what I learned about it came from the times it went quiet.

Everything else is in the archive. Everything in it is tested before it reaches you.

The lab

supercharged.sh is where I try things. Each experiment gets its own subdomain and real people using it. Most get switched off. The ones that work get built into real products.

pia, an assistant for dental and orthodontic practices, started as an experiment and is now in a live pilot. It drafts every reply and the team sends. How it was built.
Baulücken, a map of where Berlin could still build homes: empty sites, low rooftops and single-storey supermarkets. How it was built.
Custom MCP servers connect Claude to the systems a company already runs on, with everyone signed in as themselves. The first one held up, so now I build them to order. How the first one was built.

Working together

I co-founded Supercharged in Berlin with Ole Tillmann. We run AI workshops and hands-on sprints for teams: Ole works with leadership on strategy and adoption, I build the technical side, the agents and automations that keep running after the workshop ends.

Most of the teams I work with are past the demo stage. They have tried the tools, seen something promising, and now need it to work when they are not watching. If that is where you are, send me an email.

Elsewhere

LinkedIn · GitHub · X · Email · RSS