Sometimes you see somebody do something and you think: I have to try that.

But that was not how I felt the first time I saw a Rubik’s Cube.

I figured you either had the knack for it or you didn’t. Magic fingers, spatial reasoning I didn’t have, something like that. I kept that belief for years.

Then someone showed me the algorithms. There’s a method — a set of repeatable moves that work every time. It wasn’t magic. It was a process. I learned it in a weekend.

So when I came across a teenager arranging dozens of Rubik’s Cubes into a giant mosaic, I didn’t feel intimidated. I felt something closer to oh, I could do that.

She was making real-life pixel art. Because I’d solved a cube, I knew the trick she was using: you only have to get one side right. Each cube just needs to show the correct color on its face. The other five sides are irrelevant.

That’s when I ordered 100 mini Rubik’s Cubes:

One face of each cube has nine squares. 100 cubes give you 900 squares total. Arrange them into a 30×30 grid, and you have a canvas roughly the size of a large favicon — small, but enough to be recognizable if you plan it right.

The planning is where most of the work lives.

Before you touch a single cube, you take whatever you want to make — a Mario, a flower, a heart — and reduce it to a 30×30 image. Then you translate each of those 900 pixels into one of the colors a Rubik’s face can actually show. Then you map every individual cube: which of its nine squares needs to be which color, and which face needs to be facing out.

You’re writing a spec, then executing it one cube at a time.

With the Fourth of July approaching, I tried an American flag. Fifty stars, thirteen stripes. A lot of grid work and plotting before a single cube was solved.

It became the centerpiece of the table. All because I saw something and gave it a try.

Most of us in developer marketing are already experimenting with AI tools — using Claude to draft copy, maybe Nano Banana to create an image. That kind of thing.

But that’s using these tools for the work we do.

Developers are using them to build. They take something that doesn’t exist and figure out how to make it real. Sitting with an unclear prompt, iterating on something almost-right, feeling the small win when it finally works — that’s what your audience lives every day.

A lot of people take some downtime this time of year. If you get some, pick something you’d like to exist and see how close you can get with Lovable, Claude, or Cursor.

You’ll understand your developers a little better on the other side. (Something I can also help you with, BTW).

P.S. The flag is boxed in the closet. What should it become this year?