Operating principles
Honest takes from the way I actually work.
Speed got commoditized the moment Cursor and Claude went mainstream. Anyone can ship fast now, which means speed isn't the edge anymore. Taste is. You can't prompt your way to taste, you can't subscribe to it, you can't speedrun it on YouTube. It's slow, personal, built over years. It's also the one thing still doing real work.
I learned this the expensive way. The best product with no audience always loses to the worst product with one. So I build distribution into launch day, not bolt it on later. If there's no content engine ready when it ships, it's not ready.
People keep saying AI changed everything, that the gates just opened, that anyone can build now. They've been saying versions of that since computers existed. The truth is you could always have built it, you just had to learn how. AI didn't unlock building. It made the middle of the pack more crowded and the top harder to reach.
Everyone is one now, so the word stopped meaning anything. It doesn't separate good work from bad, it just lets people who don't ship feel clever about people who do. What matters is what always mattered: do you know what you're doing, and does the thing hold up when someone actually uses it.
Specialists got automated first, which would have sounded insane five years ago and is just obvious now. The people shipping real work aren't the deepest in any one stack. They're curious enough to learn six things and sharp enough to wire them together.
Beautiful, fast, and cheap used to be pick two, but that trade-off broke. With the right tools and any taste at all, you don't have to compromise on any of them. If something took six weeks and still looks generic, the work wasn't the bottleneck. The operator was.
I only build things I'd actually run myself. Every product I ship, I use in my own businesses. If I won't bet my own time on it, why would anyone else bet theirs.
The future won't come until we build it.