Most People Were Never Builders. AI Didn’t Change That.
Lowering the friction of creation exposes where vision already was; it doesn’t manufacture it. What AI changes is access, not talent — and the gap it widens.
Most kids who picked up a guitar at ten didn’t become musicians. Not because they lacked talent, but because they didn’t love the work enough to stick with it.
People who build seriously tend to share one boring trait: they’ve always been doing it, usually long before it paid, before it was cool, before it came with a title or identity attached. They tinker, redo things that already work, and don’t really leave ideas alone once something grabs them. You probably know very few people like this.
LLMs massively amplify that behaviour, but they don’t create the instinct itself. And we’ve actually run this experiment before, just in different domains.
In the 1970s, synthesizers changed who could make music. You no longer needed classical training to produce pop, and sound itself became something you could discover just by messing around. Some genuinely new and strange music came out of that shift, but the average listener didn’t suddenly stop listening and start composing.
In the early 2000s, DAWs did the same thing. GarageBand made everyone curious, and Logic Pro made a few people genuinely dangerous. Output exploded, but musical taste and vision barely moved, because the songwriters were always songwriters, with or without the means to record.
Lowering friction doesn’t create vision or creative drive. It mostly just exposes where it already was. And to be fair, it does widen access to the tools, which is a good thing, but it doesn’t flatten the curve.
That same pattern is playing out again now, but somehow the internet thinks this time is different. Everyone will build their own apps, the App Store is dead, enterprise SaaS is finished, etc. etc. Feels like wishful (or dramatised) thinking.
If you already think in systems, the leverage is enormous. If you don’t, you get a burst of motion, a few screenshots, and then you move on. This is why “everyone can build now” feels shallow to me. Anyone could always buy a guitar, and anyone could always open Logic and start producing. Ease of access never turned people into Beethoven.
What’s actually changing is access, not talent. People with good judgment but limited time or tooling suddenly move very fast. At the same time, people who relied on friction to look competent get exposed.
A lot of “experts” have really just been replaying the same local move for years. It worked, so nobody questioned it. They never had to hold the whole system in their head, and now they do. The frontend example is obvious. You could ship good-looking things for a long time without understanding backend architecture at all, and still look competent. That insulation is thinning fast.
Design isn’t the same thing as thinking in systems. You can feel when someone has taste but no real sense of structure behind it. Prompting helps you move faster, sure, but it doesn’t suddenly give you architectural intuition. Cheaper execution doesn’t magically produce judgment either — that part still takes time, or it never really shows up.
So no, I don’t think the future is everyone building their own software. It feels more like the opposite. The people who already think this way get to build much larger, more coherent systems. And there’s a widening gap between people who are good at putting parts together and those who actually design structure.