Writing
Notes on what works and what does not.
Drawn from real builds across data architecture, analytics, and AI.
The Agent Is the Model Plus the Harness
When an AI agent fails, the reflex is to reach for a better model. It is almost never the model. The capability lives in the harness you build around it, and that is the half you own.
Context Engineering Is Data Modeling
Everyone says working with AI needs a new skill. It is really data modeling, pointed at a model instead of a dashboard. If you know data, you are not behind.
What Your AI Leans On: The Data & AI Maturity Ladder
Your real AI maturity isn't the highest rung you can reach, it's the highest rung your data can hold. A five-rung ladder, and the wall it leans on.
What Ships a RAG System: The Four-Check Verification Layer
Building a RAG system is the demo. What ships it is the verification layer: four checks that decide whether to trust each answer when no one is watching.
From BI to AI: The Four-Layer Compounding Stack
Most teams treat BI, analytics, data science, and AI as separate careers. The ones that compound treat them as one stack, where each layer builds on the last.