I build AI products that work. Currently building Proxdeal, Aisbach and Multiply Content. Available for select consulting engagements.
Discover more
There is a lot of shit that can go wrong with agentic AI. But I expect that delegation will become more prevalanet with 'test-time-compute'.
I am considering moving to another country, so I had to look into the exit tax in Germany.
I built one venture, scaled another, ran an agency, and launched a podcast. Here's what I wish someone had told me last year.
AI is a tool, not a promise. I deploy it when it beats the alternatives, and skip it when it won’t.
If your LLM hallucinates or crawls, I make it production-grade.
Stop yak-shaving. Ship a roadmap your team can deliver this quarter.
Solutions that cut cost and unlock revenue—no research theatre.
I’ve battle-tested what works. For the edge cases, I tap specialists in search, data, and ML.
Get in touchThere is a lot of shit that can go wrong with agentic AI. But I expect that delegation will become more prevalanet with 'test-time-compute'.
I am considering moving to another country, so I had to look into the exit tax in Germany.
I built one venture, scaled another, ran an agency, and launched a podcast. Here's what I wish someone had told me last year.
Testing LLM preferences is 80% methodology, 20% benchmarks, and 100% trust issues.
Agents are now a marketing term, so how are we going to make it useful again?
I have a dream of a serverless MLOps platform that is easy to use, easy to understand, and easy to setup.
Don't choose a tech stack because it's trendy or because it's what the cool kids are using.