BLOG
Notes on AI, architecture, and decision-grade engineering
Practical writing on technical due diligence, vendor evaluation, delivery systems, and evidence-first AI in production.
I rebuilt my wife's vet website - to meet perfect QoE standards.
A small rebuild with a bigger lesson: in 2026, software buyers should pay for outcomes, not effort.
AI Is a Multiplier. So Is Laziness.
LLMs can speed up writing, design, and coding. They can also amplify generic output if we stop thinking critically.
AI Is Unbundling SaaS (and Changing the Economics)
LLMs make it faster and cheaper to build tailored workflows that used to require third-party SaaS. A real example: a custom call booking system built into my site.
Waste Collection Calendar: One File, One Plan, 10 minutes of prompting
A municipality PDF turns into a usable calendar in minutes: one prompt, one HTML file, and an embedded demo.
LLMs as Translators: shipping a portfolio app in about 7 hours
A one-evening case study on how LLMs compress idea to production: design kit to components, AI-assisted screenshot import, and deployment, plus the engineering judgment that still matters.
One Prompt, One Demo: Solar System Explorer
A single prompt generated this interactive solar system explorer. Here is why that matters, plus an embedded demo.
AI Nutrition Coach for Endurance Athletes
Building a personal AI assistant that synthesizes training load, wellness metrics, and sports science to deliver evidence-based daily nutrition guidance for endurance athletes.
Why RAG Beats Fine-Tuning for Most Enterprise AI Projects
For most business use cases, Retrieval Augmented Generation delivers better results faster than fine-tuning. Here's why, and when each approach makes sense.
Need a clear yes/no on the tech - or CTO-level leadership to fix it?
Book a 30-minute call. If it's a fit, I'll recommend the right starting point.
I reply within 1 business day.