Before I built my first app, I had a mental model of what app development looked like. It was wrong in almost every way. Here is what I wish someone had told me.
AI does not write apps. AI helps you build them. This distinction sounds small but it changes everything. AI is a collaborator, not a machine that takes your idea and hands you a finished app. You still need to understand what you are building, why each piece matters, and whether what AI produced is actually what you asked for.
Everything takes longer than you think. Not because AI is slow — it is extremely fast. But because building software involves a chain of decisions, and each decision creates new decisions. "Should this button be here or there?" leads to "How does this screen connect to that screen?" which leads to "Wait, does my data structure support this?" Plan for three times longer than your estimate.
Breaking things is part of the process, not a sign you are doing it wrong. Every experienced developer breaks things constantly. The difference is they have systems for recovering. For non-developers, that system is GitHub. Use it religiously.
The smallest useful version of your app is almost always better than the full version you imagined. Not just because it is easier to build — because it is easier for users to understand. One clear thing beats ten mediocre things every time.
Your prompts are your most important skill. How you communicate with AI determines the quality of what you get back. Vague prompts produce vague results. Specific prompts produce useful results. This is learnable, and it gets significantly better with practice.
Finally: you do not need to understand code to build apps. But you do need to understand what you are building well enough to describe it clearly, review what AI produces, and know when something is wrong. That is a very different skill from programming — and it is one you already have more of than you think.