Blog
Everything I've published — essays, code, demos, and half-baked ideas.
The Golden Age of PM: From Writing PRDs to Shipping Production Code
Execution is becoming the new moat. The most dangerous PM in any room is no longer the one with the best strategy deck — it's the one who can open a terminal and ship a working demo before the meeting ends.
Evals Are the New Product-Market Fit
Why a robust evaluation pipeline is the most underrated competitive moat in AI products today, and how to start building one without a research team.
Building a Minimal Eval Harness in an Afternoon
A small, opinionated repo template for getting from zero evals to a regression-tested AI feature in under four hours. Includes code.
Demo: A Multi-Agent Research Assistant
A small interactive demo I built to explore how agent-of-agents architectures can outperform single-shot prompts on open-ended research questions.
Half-Baked Idea: What if PRDs Were Prompts?
A short provocation. If a PRD is just a structured description of intent, and an LLM can read structured descriptions of intent, what happens when the PRD becomes an executable artifact?
Half-Baked Idea: The Shape of PM Craft Is Changing
Notes on which parts of the PM job are getting easier, which are getting harder, and which are quietly disappearing.