On building
The ideas were always there
For most of my career, the bottleneck was never ideas. It was access.
I've spent fifteen-plus years in operations — warehousing, manufacturing, security, tech, running my own shop. The work is always the same underneath: figure out what's broken, figure out what should exist, and make it happen. I was good at the first two. The third one usually meant getting on a tech team's calendar, building the business case, and waiting weeks for something I already knew needed to exist — because I was the one living inside the problem.
That changed in 2025.
I sat down with my oldest daughter to build a video game world with AI — half to show her this was going to matter for her, half to find out what the tools could actually do. Not what the hype claimed. What they did.
Turns out: a lot. The ideas were always there. What used to cost a budget and someone else's quarter is now just my time and how fast I'm willing to learn. That's the whole story.
Learnings
Ten months in: what building with AI actually taught me
Ten months ago I'd never shipped software. Since then I've shipped several working systems — a pattern-rendering web app, an autonomous execution engine, a couple of creative pipelines. Here's what the ten months actually taught me, minus the hype.
The hard part isn't the code — it's knowing which problem is worth solving. AI will happily build you the wrong thing, beautifully. Judgment is still the scarce resource, and judgment is the thing fifteen years of operations actually bought me.
Shipped beats perfect, but shipped means finished. Not half a feature with a confident demo. A demo proves it's possible; production proves it's real. The gap between those two is where most of the work lives.
The tools change weekly. What was true in the spring was stale by summer. Keeping up isn't a phase you finish — it's the job now. That's uncomfortable, and it's also the opportunity.
Most of the gatekeeping was financial, not technical. What used to require a team and a budget now requires time and willingness. That's a bigger deal than it sounds — because if it's true for me, a non-technical operator, it's true for a lot of people who don't know it yet.
And it's real. Not chatbots, not autocomplete. A genuine force multiplier for anyone with good ideas and the judgment to point it somewhere worth going. That last one is why I started writing this down.
On the lane
What operators see that engineers miss
I'm not competing with engineers. I'd lose. I'm working in a lane most of them don't occupy.
Engineers are trained to build the thing right. Operators are trained to know which thing to build — and, just as important, which thing to quietly kill. When you've spent years inside the messy middle of a business, you develop a nose for the difference between a problem worth solving and one that just looks interesting.
AI tilts the field toward that instinct. When building gets cheap, the constraint stops being "can we build it" and becomes "should we — and what actually happens when real people use it on a Tuesday." That second question is an operations question. It always was.
The most valuable AI practitioners coming out of this era won't all be the best coders. A lot of them will be people who deeply understand the problem AI is being pointed at, and who now, finally, have the tools to point it themselves.
That's the bet I'm making on myself. So far it's paying off.