Field notes, thought experiments, opinions, and insights on building with AI.
Notes that blend product, engineering, business, and design into one working craft.
LLMs eat the scaffolding: Context is next
Why context engineering, retrieval, and tool selection are becoming the next critical scaffolding layer for AI agents before those capabilities get absorbed into the model stack.
Modern Artisans: The Future of Work
A short argument that the future of work may belong to modern artisans: independent, high-agency knowledge workers using AI to own outcomes end to end.
Pair programming with agents
A lightweight workflow for using coding agents as a thought partner: tighten conventions first, then iterate in public with your README as the source of truth.
How humans manage deep contexts & why agents are no different
Why Contextrie exists: how humans compress context, why agents rot, and what a Chief of Staff pipeline looks like.