— Notes
Short, opinionated notes on what actually breaks
MarTech architecture, AI engineering, and growth systems — written from production incidents, not theory.
RAG Is Not a Silver Bullet: Where Retrieval Actually Breaks Down
RAG gets pitched as the fix for hallucination. In production it has its own failure modes — and most teams don't find out until the demo is over.
The MarTech Stack Nobody Needs: A Field Guide to Over-Engineering
I've audited stacks with six tools doing the job of two, each justified individually and indefensible together. Here's how that happens and how to catch it.
I Stopped Reviewing AI-Generated Code Like It's Human-Written
Human code fails loudly in ways you've learned to spot. AI-generated code fails quietly, confidently, and in a completely different shape — so I changed how I review it.
Server-Side Tagging Isn't About Privacy — It's About Owning Your Data Pipeline
The pitch for server-side GTM is almost always framed as a privacy or ad-blocker workaround. That undersells what it actually gives you: control.
Why Most GA4 Migrations Fail (And It's Not a GA4 Problem)
Teams blame GA4 for broken reports, but the real failure almost always happened before a single tag was configured — in event taxonomy nobody agreed on.