Is building an app with AI difficult? Yes and no. Getting started can be quick. Returning to the part that still does not work is where persistence begins.
Generating a first version is a beginning. Before someone else can rely on it, you need to understand what it does and check where it fails.
Describe a small thing the app should let a person do. What should happen when they try it? What would tell you it worked? Keep that question narrow enough to examine in one session.
When the result is different from what you expected, write down the difference. An exact description of the problem gives you something to work on. Asking for an entire app to be made better is much harder to judge.
Return to the feature and try it yourself. Follow the path a new user would take. If you cannot explain what you have built or how it behaves, make understanding that your next task.
Progress can include removing something confusing, fixing a broken step or deciding an idea needs a different approach. The point of my reply was to keep working and learning. Persistence is more useful when each return to the project has a clear question attached.