Essays on AI investing and building.
Four pillars: AI investing, building with AI, founders and scaling, and determination. New essays as the work surfaces them. No jargon parade, no framework slop, no reheated VC takes.
- Jul 11, 2026·AI Investing·7 min read
Why I pass on great teams
The hardest passes have nothing to do with the founders. The team is the easy read. The shape of the market they are standing in is the job.
Read essay → - Jul 11, 2026·Building with AI·7 min read
Why I kill agents that work
The dangerous agents in my stack are not the ones that fail. Failure announces itself. The ones that work go quiet, and quiet is where the risk accumulates.
Read essay → - Jul 11, 2026·Determination·7 min read
The quarter is the wrong unit of progress
A twelve-month training block taught me that the work that compounds is invisible at ninety days. Most companies are managed at exactly the resolution that hides it.
Read essay → - Jul 11, 2026·Building with AI·8 min read
For an AI agent, the eval set is the codebase
I stopped treating evals as a testing chore and started treating them as the thing I actually own. Everything got easier the day I made that switch.
Read essay → - Jul 11, 2026·AI Investing·7 min read
AI revenue is not SaaS revenue
The ARR slide looks like software. The cost of serving it often does not, and the gap between those two numbers is where I spend my diligence time now.
Read essay → - Jul 8, 2026·Founders & Scaling·7 min read
The headcount trap: why I want my AI founders to stay small
Every founder I back eventually asks me when to hire ahead of the curve. My honest answer is later than you think, and sometimes never.
Read essay → - Jul 4, 2026·AI Investing·7 min read
I diligence AI companies from the terminal, not the deck
The deck tells me what the founder wants me to believe. The product, the repo, and the logs tell me what is actually true.
Read essay → - Apr 10, 2026·AI Investing·7 min read
Why I'm passing on AI wrappers in 2026
The moat question is louder than it's ever been, and the answer most founders give is the wrong one.
Read essay → - Mar 28, 2026·Building with AI·9 min read
My production AI agent stack, six months in
What I run, where it breaks, and the unglamorous glue-code that determines whether any of this actually ships.
Read essay → - Mar 14, 2026·Founders & Scaling·6 min read
What I tell AI founders about distribution
Model quality is the easy part. Getting it in front of users who will pay for it is not.
Read essay → - Feb 28, 2026·Determination·5 min read
Training plans and investment theses have the same shape
Both are bets on compounding discipline across a twelve-month horizon. Both break in the same ways.
Read essay →