I shipped three apps and co-founded a fourth without writing a line of production code myself. Two years ago I never could've imagined saying that. Today it's just the beginning. But I want to be careful about what that sentence actually claims, because it's easy to conflate two things that are not the same: a new capability and a new skill.
A skill is learned and earned. You get worse before you get better. You build a mental model of the terrain over months or years, and somewhere in there you develop the instinct to catch your own mistakes before anyone else has to. A capability is different — it's about output, not about you. AI gave me the capability to ship a working app in a week. It did not make me a developer. Those are two different claims, and treating them as the same one is where I see people — myself included, in weaker moments — get into trouble.
I felt this gap directly the first time I checked my usage against the top users at my company. I'd been in these tools daily for months and assumed I was ahead of the curve. I wasn't close. I reached out to the person at the top of the list and asked how he worked. His answer was simple: use it for everything, not just the obvious tasks, and don't assume you know its limits, because the limit moves every week. That reframed the tool for me — not as a designer and code generator, but as a fast collaborator that still needs someone deciding what "good" means. I'm that person for design, user experience, and research. But I understand my limits in coding and product marketing better than ever.
That's the part capability can't do for you. AI will hand you an interface, a flow, a first draft of anything — with total confidence, whether or not it's right. It doesn't know the difference between a decision that's defensible and one that's merely fluent. That's still a judgment call, and judgment isn't something a tool grants you. It's the thing twenty years of shipping products taught me, and it's the reason I can look at AI-generated output and know something's off before I can even articulate why in technical terms.
So here's where I've landed: the output AI gives me is not the deliverable. It's a fast first draft that still needs an expert pass before it earns the right to ship — a quality check, a judgment call, someone asking "should we" and "why would we" instead of just "can we." Skip that step and you're not moving faster, you're just moving the risk downstream to whoever discovers the gap later.
I'm not becoming a developer, and I've stopped pretending the apps I ship make me one. What I've actually become is someone who can turn ideas into working software faster than ever — because I know exactly which parts of that process still require the person who put in the twenty years. AI closed the distance between having an idea and seeing it exist. It didn't close the distance between fluent and right. That part's still on me.