Joseph Simmons

The shop

About forty people. Screen printing, embroidery — including puff and multi-location work — heat transfer, laser etching, and contract decoration for brands that ship under their own label. We're a SanMar PSST decorator, which means blanks arrive from several warehouses and leave as one shipment, and if you've never had to reconcile that against a purchase order at 6pm on a Thursday, count yourself lucky.

It's a real business with real constraints. Nothing in the book was done in a lab or on a weekend project. Every change had to survive Monday.

The software history

Impress ran the shop until 2022 and was, for a long time, fine — until the industry moved and it didn't. Then came NetSuite, which lasted five months. Then ShopWorks, for about three years, mostly out of exhaustion.

The NetSuite one is why I wrote a book instead of just switching quietly. It wasn't that the software couldn't do things — it could do almost anything, as long as I paid someone to explain my own shop to them first. Every change I wanted meant a scoping conversation with a person who had never seen a press. By the time they understood the problem well enough to quote it, the quote cost more than the problem.

What I eventually noticed is that this is the shape of the deal, not the failing of one vendor. You rent a system built for the average of ten thousand businesses, and then you spend the rest of the contract paying — in labor, by your own people — for the distance between that average and you. Nobody puts that line on the invoice.

Then Odoo, which I'll be openly enthusiastic about anywhere you let me. It's the first one where the answer to 'can we change this?' was just yes.

Why I wrote it

Because after the NetSuite implementation collapsed I decided I wanted to help other owners get out from under systems like it. And because in the middle of building our own with AI, I kept catching myself thinking somebody should have told me this. Not the technical parts — those are documented. The judgment parts. When to configure and when to build. How to describe your own process precisely enough that it can be turned into software. What to do when the AI confidently writes something wrong. How to know a project has quietly become a second full-time job.

So the book is what I'd have wanted on day one: one shop's real build, the framework underneath it, and the failures left in.

The AI part, stated plainly

I built this with AI assistants doing the code and me doing the deciding. Claude is named in the book because it's what runs in our development workflow and what's connected to our system directly; we also use ChatGPT inside Odoo as a knowledge base. I have no relationship with either company. If a better tool shows up next year, I'll use that one, and the method in the book won't change.

Talk to me

If you run a decorating shop, or you're partway into a build of your own and stuck, email me at joe@superiorimprints.com. I read everything. I answer most of it.