I love technology. I’ve been using it my whole life.
I’m part of that odd generation that got to experience both sides of the digital divide. As a kid we had three or four TV channels. By the time I was a teenager the internet arrived and suddenly I had access to almost everything and could stay connected with friends. In college I was a sophomore the year Facebook launched. I still remember my friends coming back to the dorm saying, “You have to join Facebook. If enough of us request it, they’ll add our school.” We did, and now it’s just a normal part of life. Weird how that works.
That early exposure stuck with me. In my twenties I played with Arduino boards, Raspberry Pis, little stepper motors, and kits. I built the usual projects, copied code from forums, and eventually ended up with a Rubbermaid container full of parts in the basement. I even tried getting my son interested years later. It never really stuck for him.
Then a real shop problem showed up.
The Problem That Forced the Decision
We regularly needed weld overlays on shafts. Nothing exotic—just a controlled, consistent overlay with limited heat input. I knew a shop that could do it, but they always took six to eight weeks. For the kind of work we do, that wait was painful. One day a regular customer asked if we could handle it. I said yes.
I walked straight into one of my engineers’ offices (he’s an excited twenty-something) and told him we were going to build a linear welding machine. The cheapest commercial package unit I could find was around twenty thousand dollars. We decided to replicate the core idea with good mechanical know-how, an Arduino-style board, and a handful of parts.
Building the Mechanical Side
We spent about a day and a half picking a stepper motor and driver combo, choosing linear rails, and ordering a torch-head holder. My engineer laid everything out, bought the travel rail and Acme screw, and paired it with the motor. Because we have a full machine shop and a senior machinist with forty years of experience, the custom couplings and mounting pieces were made in-house and fit perfectly. That advantage is huge—trying the same build in a garage would have been a lot harder.
In roughly a week and a half the mechanical assembly was together and moving the way we wanted.
The Coding Part That Used to Be the Hard Part
Then came the software. In the old days this is where projects usually died for me—hours of copy-pasting, debugging, and staring at pinouts. This time I opened Grok, dropped in the Amazon links for every part I had bought (board, LED display, control knob, switches, stepper driver), and simply described what I wanted the machine to do.
Within about a minute I had clean, working code. I uploaded it, plugged everything in, and it did exactly what I asked. A few small tweaks later and it was solid. Grok even gave me the wiring terminations and a clear map for the board. The little LED display came up showing the numbers I needed without me having to spell out every detail.
The hardest part of the entire project—the programming—had turned into a couple of text messages.
Making It Actually Weld
The following week we moved the rig into an unused booth, paired it with a positioner, and hooked it to an old welder. My engineer ran the first sample while I was tied up elsewhere. He brought it in and said it looked terrible. Fair enough—settings needed work.
I’m not a production welder, but I understand the variables. I grabbed my hood, and I dialed in travel speed, heat, and positioner rotation. After a few adjustments the puddle looked right and the sound was right. We let it run several rotations, shut it down, and looked at the result.
It was dead-on. The kind of clean, consistent overlay people pay serious money for.
We repeated the settings a few more times on the same diameter to lock them in, then threw an actual customer shaft on the machine. Same result. Suddenly we had a capable in-house overlay welding rig.
Total parts cost was around three hundred dollars plus a week of back-burner work. My foreman—who never gives anything away for free—looked at the weld and simply said, “Damn, that looks good.”
What This Really Means
The mechanical work still takes skill, tools, and time. Having a real machine shop made that part realistic. But the technology barrier that used to stop people cold is almost gone. AI didn’t just help with the code—it removed the intimidation factor. I no longer have to be a programming expert to make hardware do useful work.
I use the same approach for everything else now: writing company procedures, drafting difficult customer letters, thinking through processes. What used to take a full day of careful wording can be refined in minutes with a second set of eyes that never gets tired.
Technology has always been my friend. AI just made the friendship a lot more productive. The mountain that used to look too steep is now something you can walk over in an afternoon if you’re willing to try.
And once you’ve done it once, you start looking for the next problem you can solve the same way.