Background in strategy consulting, private equity, M&A and investment banking. Since April 2026 building AI systems and agents for finance and operations at a stealth AI lab in Berlin, San Francisco and Amsterdam.
It was 11 p.m., and the deal team needed 200 emails to potential buyers by the next morning, each with the same text, written one by one. I had the solution in my head: a short script would have done it in minutes. Unfortunately that was not the standard process, so I spent the night copying and pasting. Since then I have followed one rule: never accept "we have always done it this way" as a reason. For every task I look at all the options, from existing tools to new technology to something I build myself, and take the one that gets to the goal best.
A few months later I had signed a permanent contract in M&A. Then a stealth AI lab asked me to build its finance from the ground up and set the company up to run on AI. It was a big risk, but a chance like that comes once in a lifetime, and I took it. The boss I was about to join understood what an entrepreneurial opportunity this was, wished me every success and told me to get in touch any time. Finally I could try any technology I wanted and build the environment I wanted to work in.
Since April I have built a private knowledge system on open-source models, an agentic CRM for the funding round and a finance stack that checks itself every night. What I am proudest of is the company brain. Now I am looking for the place where I can build like this for years.
Company knowledge sat across Notion, Drive, Slack, meeting transcripts and the fundraising database, and none of it was allowed to pass through an outside vendor. I built a retrieval system that answers with the source line, runs entirely on one machine with open models, and refuses when the answer is not in the knowledge base. Next to it, a personal second brain that my cockpit feeds every night.
Open the caseA pre-seed round, no tool that fit. I built a full portal: a Kanban pipeline, a daily action queue, a scored investor list, the team's combined network and a news feed, all fed every morning by an engine that reads Slack, meeting notes, calendars and the web. The agent proposes, the team decides, the agent never sends.
Open the caseA client saw its numbers through accounting reports that arrived two months late. I built the data pipeline, a dashboard and a forecast, then recalculated every figure against the raw files.
Open the caseA stealth AI lab ran on a dozen tools at once. I built one local app that answers the daily questions in one place: did everything run overnight, who read the data room, what needs a decision today, and are the books right.
Open the case