I asked AI to create Visual Programming Components, a task that takes months of learning.

And it created perfect beautiful things, but only because I knew what to ask for.

In fact, I did not mention, that I wanted to build a Visual Programming Language.

I asked for a line, in HTML, as a stand-one web component, similar to a button or an input box.

I also told it to create a window component, but this time, I gave it unfinished code as a prompt.

And I finally asked for a Signal based web page element monitor, that with return x, y, width and height, also inducing a tiny bit of code.

If I just asked for visual programming components, I am sure it would have created something fantastic.

But there is no guarantee, it would work in the long run.

The reason I knew what to ask for, is because this is the best case scenario based on my research.

In fact, it is frightening, how well the AI responded, and how it fixed a small oversight in the code it generated.

It is of some note that, I asked that WebComponents use Signals, to keep the code clean, the code is nice.

I attach a demo of the components, resizable/draggable windows and port position tracking just a couple of minutes to make.

But the first six versions of so I wrote by hand, not only is that a good way to discover the best solution.

But also update your knowledge, or just practice and learn JavaScript.

Creating multiple versions of your program, will also reveal what is unimportant.

I originally intended to support Pan and Zoom, I thought that using the idea of scroll bars on a web page is primitive.

But I was wrong, because adding more than the basics, entails several other things and results in explosion of complexity.

You lose control over the code, as there are too many things to remember.

This approach that I have here, is just two HTML tags window-container and flow-connector.

They are so generic, that many visual programming languages and interfaces can be created.

Finally, this small programming project begun this afternoon, as a result of several problems I discovered in Visual Programming.

The biggest one being is that Visual Programming Languages, do not capture the architecture of the entire application.

But also that they are slow for programmers, and when connecting one output to many inputs…

The order of operations is undefined, you don’t know where data is going to go next.

For sequence control you have to connect everything in a series, one after another, that does not simply things.

That is an unnecessary restriction, you should connect your boxes any which way you want.

Visual programming is about simplicity, not restriction and rules you have to remember.

The AI allows for creating dozens of experiments, in the span of a week.

It helps to uncover the correct solution, by experimentation, and quick turn around.

Not only does it create awesome code, it also helps you learn how to make it better.

I thing the next step in visual programming, is free form diagramming.

A thing that allows to draw out a system diagram, with no rules at all.

This diagram serves as a map, and a table of contents.

That helps programmers and the novice, finally, finally, visually understand architecture.

Without AI I could not afford this research as it would take months, with it, it is almost already done.