AI wrote my code… now what?

I tried building something with AI recently, the kind of thing that would have taken me hours. I typed out what I wanted, hit enter, and watched as it stitched together everything, almost instantly.

It worked on the first go, which felt both exciting and slightly unsettling. For a moment, it genuinely made me wonder if building things had always been this simple and I had just been doing it the hard way.

So I leaned into it.

Every time I needed something, I asked AI. A new feature? Done. A bug? Fixed. An improvement? Suggested and implemented before I could even think it through. At some point, it stopped feeling like I was building anything at all and started feeling more like I was assembling pieces someone else had already figured out.

And then came the smallest change.

Nothing dramatic—just a minor tweak. But suddenly, I didn’t know where to start. I couldn’t tell which part of the code was responsible, what depended on what, or why something had been written the way it was. AI could still generate answers, of course, but I didn’t even know what question to ask anymore.

That’s when it clicked.

Using AI without understanding what it’s giving you is a lot like copying homework. It works beautifully as long as no one asks you to explain it. The moment you have to make a change, or adapt something slightly, you realize you never really understood the path that got you to the answer.

The issue isn’t AI—it’s how we use it.

Because AI is an incredibly powerful tool. It can speed you up, help you experiment, and make building things more accessible than ever before. But it doesn’t replace your role in the process.

The difference comes down to this: are you using AI to avoid thinking, or to think better?

So, what should you do differently?

If you’re going to use AI to build something, don’t make your first step the prompt.

Say you want to build a full-stack website for your company. Before asking AI to build it, spend some time understanding what you’re actually asking it to create.

Start with the basics:

What is the frontend?

What is the backend?

What is a database?

Then go one layer deeper. Understand things like authentication, APIs, data flow, hosting, and deployment—not so you can build all of these yourself, but so you have a mental model of the system AI is building for you.

Think about it this way.

If AI tells you, “We should move this logic to the backend because it shouldn't be exposed on the client,” would you know what that means?

You don't need to know exactly how to make the change. You just need enough understanding to ask, “Why?”

And that's where your interaction with AI starts to change.

Instead of:

“Build me a website with a login system.”

you can start thinking:

“I need a frontend where users can log in, an authentication flow on the backend, and somewhere for user credentials and account information to be stored. How should these pieces fit together?”

The difference isn't that the second prompt contains some magical combination of words.It's that you understand the problem better.

And that means you can question AI's decisions, spot when something doesn't make sense, and make better decisions about what should happen next.

So before your next AI-built project, try this:

1. Understand the system you're about to build.
Learn the basic concepts and terminology.

2. Break the problem into parts.
What needs to happen on the frontend? What belongs in the backend? What data needs to be stored?

3. Ask AI to explain the architecture before asking it to build everything.
Ask what it recommends, why it recommends it, and what the trade-offs are.

4. Build one piece at a time.
Let AI build, explain, and test each part as you go instead of generating an entire application and hoping it works.

5. When something breaks, understand the problem before asking AI to fix it.
Even a rough idea of where the problem lives can completely change the quality of your next prompt.

The goal isn't to know everything AI knows, but is to know enough to stay in the driver's seat.And once you have that basic understanding, there's another piece of the puzzle:

How do you actually communicate what you want AI to build?

Because knowing what you want is one thing. Knowing how to structure your prompt so that AI can understand the problem, reason about it, and build it correctly is another.

That's where we go next.

Stay tuned—I’ll be sharing a blog on how to structure your prompts for better results soon.