The work
Systems we’ve built and shipped.
Real projects — products, internal tools, AI agents, infrastructure — with notes on what we built, why, and what we learned. Updated as work ships.
Human on the Loop
AI can write most of the code now. That was never the hard part. How we built a software team of specialised AI agents with real separation of duties and enforced quality gates, and what changes when the human stops approving every step and starts supervising the line.
↗ Learn more
The Personal Health Concierge
Concierge medicine sells depth it cannot afford to give. We built an AI employee with a panel of one patient, and it found a pattern four different specialists each held one piece of. A case study in using AI to differentiate an offering and widen margins at the same time.
↗ Learn more
The Anatomy of an AI Employee
How to build agents that work like colleagues, and manage them like a workforce. The five layers of an AI employee, the two seams that matter, and what to own versus what to rent.
↗ Learn more
MungerMind: An AI Analyst That Reads the Whole Market
A good analyst can carry a few dozen names in real depth. The S&P 500 and Russell 2000 hold around 2,500. MungerMind reads the entire universe against a single thesis in hours, then surfaces the two or three names worth deep human work.
↗ Learn more
The Harness is All You Need
Today’s models are already capable enough to do most of the work that happens at a computer. What stands between a clever chatbot and a dependable digital employee isn’t a smarter brain. It’s the harness around it.
↗ Learn more
How to Think About Governing AI in Production
AI risk doesn’t arrive when you adopt AI. It arrives the moment a model stops assisting a person and starts doing the work. A simple, proportionate way to think about governing it, plus a free field guide.
↗ Learn more
