Build the Expertise Into Each Folder
How I scope AI coding agents to the filesystem: each folder carries its own memory, skills, and conventions, many agents write to one shared log, and the whole setup travels to whoever opens it.
Writing
Writing
Longer pieces on AI and automation in small-manufacturing operations, organized across four pillars: the problem, the methodology, the implementation, and the results.
- The ProblemWhy SME manufacturing AI is different, and why closing the gap matters.
- The MethodologyThe Manufacturing Intelligence Framework: two interlocking frameworks, eight phases, one feedback triangle.
- The ImplementationHow the systems actually get built: tools, agents, what broke.
- The ResultsQuantified before/after outcomes from real deployments.
The Backbone Without the Tools
Small manufacturers are most of US manufacturing, yet they are the ones locked out of the operational AI that large firms use. Why that gap is a national problem, with the numbers.
Catalog Automation: From Weeks to One Day Across Ten Product Lines
A product catalog kept by hand is stale the day you finish it. Here is how to stop maintaining one and generate it from live data instead, and what that did to a refresh that took weeks.
Why SME Manufacturers Can't Just Use Enterprise AI Tools
The story is that AI will help small manufacturers catch up by buying the tools. But enterprise AI assumes a budget, clean data, and a team SMEs do not have. Here is what works instead.
Why You Can't Just Ask AI for a Quote
Hand a model your prices and a takeoff list and it returns a beautiful quote that is quietly, badly wrong. A model is only as good as the harness around it. Here is what a real quote needs.
What I Built Inside a Live Company
A companion to a more personal essay: not what it feels like to build inside a running company, but what the constraint produced: the system, the decisions behind it, and the lessons.
Building Inside a Live Company
What it is actually like to build software inside a company that is already running, for people who never signed up to test it, and why the constraint I kept resenting turned out to be the point.
Every Business Sits on a Product
Decompose the product to its atoms and everything above it (pricing, quoting, sales, marketing) becomes a calculation, not a guess. And phases 5–7 are a loop, not a line.
The Shape of the System
Not a feature list but a structure: one engine, with quoting, sales, factory costing, production data, an agentic layer, and customer tools all built on it.
One Engine, Two Audiences: Letting Customers Quote Without Seeing Costs
How to put a customer portal on top of an internal engine that knows every cost: reuse the engine, and make the boundary structurally unable to leak instead of trusting yourself to strip fields.
What I’m Documenting Here
A short orientation to the writing: what it covers, how it’s organized around a phased framework, and where things currently stand.