Grant

There's a lot of ways that you can improve your like initial prompt in order to,

create something that's going to have the best shot on goal in order to win.

There's also a lot of ways that you could potentially improve your skills by reading other developers who are experienced in this space.

And they're actually trying to figure out what is the best way to code with AI in this way.

And also just...

kind of try to teach yourself on the fly.

Like let's say, try to build something really, really simple and then go back once it's done and it's working, ask Gemini, okay, explain exactly how this works to me and show me step by step, explain and break it down.

And then go through the code and then ask like, okay, for this piece of code, what does this code do?

How does it work?

You can even ask it to do this, let's say you're working in O3 Pro or Gemini 2.5 Pro in the chat,

model, you could ask it to help you design a program from scratch and all the features you need before you ever take it to code.

So like say, okay, this is what I want to do.

Let's break down feature by feature what I need to build in order to do this, and then work with it backwards to reverse engineer it.

So you're not coming up with

one sentence prompt, take it to Google Builder and expect it to build you a full application.

But instead you are, you know, working incrementally in order to program something.

I think that's the way I would go about it if I were.

Uh, same, same issues over here.

I'm letting it, uh, try and work itself out.

Looks like here it says, to provide a rich preview for the user interface without needing to make a live API call, I'll populate the application with some realistic mock data.