Writing · Nº 0456 essays
The writing
Notes on building AI companies, operating a portfolio of one, and the playbook for the AI era.
ThesisGovernance is the productEveryone is racing to make AI do more. The harder, more valuable problem is making it do exactly what you asked and nothing you didn't. That is what I actually sell.OperatingWhy I build twenty companies at onceConventional advice says focus on one thing. I'm doing the opposite, and it only works because of one decision I made about the foundation underneath all of them.EngineeringThe operator's stackHow one person ships and runs a portfolio of AI products without a team. The boring infrastructure decisions matter more than the AI.OperatingI host it myselfRenting your whole stack from a managed cloud is renting your own business. I run the portfolio on hardware I control, on purpose.OperatingPutting my face on itFor years I built quietly and let the products speak. Now my name is on the door. Here is why I stopped hiding behind the work.ThesisWhat AI-native actually meansMost products bolt a chat box onto an old app and call it AI. AI-native is the opposite. The model is the mechanism, not a feature.StrategyAI Native Beats AI Bolted OnProducts where the model is the mechanism will outcompete products that staple AI onto an existing workflow. A competitive prediction, not a definition.StrategyGovernance Is the Moat Nobody Wants to BuildClaims discipline, consent records, audit trails, and human checkpoints are unglamorous. That is exactly why they make a durable competitive moat.StrategyThe Economics of Shipping Twenty CompaniesWhen the foundation already carries auth, billing, infra, and governance, the marginal cost of the next venture collapses. The math behind a portfolio.ThesisOne Governed Foundation Under Every CompanyNineteen ventures share one governed substrate. Identity, audit trails, and guardrails are built once and inherited everywhere, not rebuilt per app.EngineeringHow to Design a Product Around a ModelProduct design when the model is the core mechanism: shape the workflow around what it does reliably, design for non-deterministic output, and build the recovery paths.OperatingHow to Prove AI Reliability to Enterprise BuyersWhat enterprise buyers need to sign: evidence the system will not lie and will not do damage, packaged as evals, audit logs, and claims you can defend.