Last year I wrote about offering my brother an AI system for his sign shop and getting the fastest no I’ve ever heard: we don’t like AI here.
I’ve thought about that answer a lot since, because he wasn’t being a luddite. He was telling me I’d shown up with a solution wearing a costume. What he actually wanted was for a quote to stop eating forty minutes of his afternoon, and for someone to tell him whether the last one made any money. Nothing about that requires the word “AI” to appear anywhere.
So we stopped pitching and built the software. It’s called PrinterFlo, it runs print and sign shops end to end, and there’s a full write-up of the build here. This post is the other half — what a studio that mostly builds websites learns by shipping an actual product.
The interface was never the hard part
We are good at the visible layer. That’s the job. I assumed the rest would be a matter of grinding through it.
The rest is where the year went. Taking a payment is easy; taking a payment into the shop’s own account so we never touch their money, never hold it, and never become a party to their chargebacks — that’s a different piece of engineering entirely, and it’s the only version a shop owner should accept. Pushing an invoice to accounting is easy; making sure the invoice number, the tax code and the customer record all match what’s already in their books, forever, is not.
None of that is visible in a screenshot. All of it is the reason people either trust software with their business or quietly keep using the spreadsheet.
If you’re weighing custom software for your own business, that’s the question worth asking whoever’s quoting you. Not “can you build the screens” — everyone can build the screens. Ask what happens to the money, and what happens to your data if you leave.
Owning the thing changes what you’ll accept
We already told clients not to rent their business from a platform. Building a product made that concrete in a way the argument never did.
When you run the software yourself, every shortcut is a bill you personally pay later. A slow page isn’t a score in a report, it’s a shop owner on a job site with two bars of signal giving up. Data you can’t export isn’t a policy detail, it’s the reason someone won’t sign up in the first place. We shipped a one-click full export early, on purpose, because a product that’s hard to leave is a product that has to be worth staying in.
That posture came straight back into the client work. It’s why the sites we build stay static and fast, why the client owns the repository, and why there’s no monthly fee to keep your own website online.
Distribution is half the build, and nobody warns you
The part I was least prepared for: finishing the software gets you nothing.
So we built the demand side too — comparison pages, guides, and a set of genuinely free tools that shops can use whether or not they ever pay us: a channel letter designer that exports real cut files, a price calculator, a print file checker, an image converter I now use on client projects weekly. We published a plugin to the WordPress directory. We started a forum.
The first shop that ever paid us found the product through a Google search, not through any of the outreach we’d been doing. We checked — the search data against the outbound lists — because it would have been easy and wrong to credit the cold email.
That’s the same lead engine we build for clients. Building it for ourselves first meant we got to be wrong about it on our own time.
What it means if you’re the small business
Most shops don’t need custom software. Off-the-shelf is cheaper and faster and usually correct.
It stops being correct at a specific moment: when the way your business actually works has become a pile of workarounds inside a tool that was built for somebody else’s business. Ten browser tabs. A spreadsheet that only one person understands. A whiteboard that is the real system of record and lives in one room. That’s not disorganisation — it’s a product that doesn’t fit, and everyone patching around it every day.
That’s the point where building the thing pays for itself, and it’s the work we like most. If that sounds like your operation, tell us about it.
And if you happen to run a print or sign shop, the software already exists. My brother was right about the AI. He was right about the forty minutes, too.