One problem
I’ve never liked solving the same problem twice. It’s honestly what got me into computer science: why solve something once when you can build the thing that solves it every time after?
That’s still how I think about AI. For a while I used it for whatever was in front of me. I’d ask, get a good answer and move on. Which is great. It’s just the smallest version of what’s possible.
A class of problems
The catch was always cost. Building the mechanism took real time, so I only bothered for the big, obvious problems. Everything smaller, I just kept solving by hand.
AI changed that. Now a mechanism is cheap enough to build for almost anything. And once I have a few, they start feeding each other: one answers the question, another notices which questions keep coming up, another fixes whatever’s causing them.
I figured each fix would just get easier. I didn’t expect them to add up. Ten small fixes used to stay ten small fixes. Now they turn into one bigger thing that can take on problems none of them could alone.
The system
Here’s the kind of thing I mean. Most suggestion boxes are where ideas go to die. Someone writes “we should fix onboarding,” and it just sits there, because nobody knows what it actually means.
Now picture a box that talks back. It asks the follow-up questions a good manager would, until the idea is clear, then sorts it: worth keeping an eye on, or ready to build. Nobody has to chase anybody. I’ve built a version of this.
And it’s only the first layer. Once ideas show up clear, it changes how work moves. That changes how things get built, and eventually how the whole place runs.
Everyone
Every layer I build is aimed at the same thing: the person closest to a problem should be able to solve it themselves.
So I build the mechanism, then hand it over. Someone describes what they need, the system draws out the details, and something real gets built without waiting on a specialist. Each time, a problem gets solved for good by the person who actually had it.
That’s what I care about most: not the hardest problem I can solve, but how many people get to solve their own.