For roughly fourteen years I wrote code and consulted for companies trying to figure out what their systems should actually do. Then, five and a half years ago, I took over a metal fab and machine shop. The two stretches feel like they happened to different people, and that's the most useful thing that's ever happened to me professionally.
Good developers and consultants try to learn the business they're working in. The serious ones spend real time on the floor, ask good questions, and build genuine domain understanding. I tried to be one of those. But nothing — nothing — compares to signing the front of the checks, owning the outcomes, and living inside the business day after day. You don't really know an industry until your livelihood depends on getting it right.
A few things I didn't fully understand until I lived them:
A quote isn't a form. It's a phone call, a napkin sketch, a text message with a blurry photo, a follow-up email, a walk out to the shop floor to ask the guy who ran the last similar job what it took, and then a spreadsheet that someone built in 2014 that nobody is allowed to modify but it must be used. The "quoting bottleneck" isn't a UX problem. It's an information-archaeology problem.
Tribal knowledge is real, and it is load-bearing. The most valuable person in a small shop is rarely the one with the title. It's the person who remembers that this particular customer always wants the holes deburred a certain way, that this alloy work-hardens if you push the feed, that the second op on this part is easier if you flip the fixture. None of that is written down anywhere. If that person retires, you lose a quiet percentage of your margin and you don't even know which percentage.
Operators context-switch constantly, and every switch has a cost the software world tends to underestimate. A machinist might touch a router, a caliper, a tablet, a paper traveler, a whiteboard, and a phone in the span of ten minutes. Each handoff is a small tax. Add them up across a shift, a week, a year, and you're looking at real money, money that never shows up on a P&L because nobody knows how to categorize it.
The systems that exist for shops like mine were mostly designed by people who'd never run one. They assume clean inputs, disciplined data entry, and operators who have time to babysit a UI. The actual environment is loud, dirty, interrupted, and optimized for getting parts out the door. Software that demands attention loses to software that earns it, and most of the legacy stuff doesn't earn it.
Paperwork is the silent operating cost of small manufacturing. Certs, travelers, inspection reports, packing slips, customer portals that each want their own format. A two-person office can spend a third of its week just moving paper around. None of that paper makes a part.
And yet, and this is the part I want to be honest about, I'm coming back to software more optimistic than I left it. Because the gap between what's possible and what's deployed in industrial businesses right now is the widest I've ever seen it. Not because the technology is magical, but because the baseline is so manual that even modest, well-designed tooling produces outsized results. A practical workflow system that just captures tribal knowledge as it happens, or quietly pre-fills a quote from prior jobs, or turns a photo of a part into the start of a router. These aren't sci-fi. They're achievable now, with components that already exist, if someone designs them for how shops actually work instead of how a software roadmap wishes they worked.
So I'm pointing myself back toward systems and tools, but with a different set of assumptions than I had when I was last writing code full time. I want to spend the next chapter building things for operators: practical workflow systems, sensible automation, AI applied where it earns its keep and left alone where it doesn't. I'll be writing about what I find as I go — what works on the floor, what doesn't, what the vendors get wrong, and where the real opportunities are hiding in plain sight.
Maybe that eventually leads to building a new kind of highly efficient, modern shop from the ground up. If it does, it'll be built with a very different understanding of where the real leverage lives.
The interesting work is in the boring parts, and there's still a lot of boring parts left.