Pizza Fractions and the AI-Assisted Dad
My son asked me a question on the way home from school today: “What does the one slash two mean? 1/2?”
And I just – lit up. That’s the only way I can describe it. One of those moments where a kid asks something so simple and fundamental that your brain immediately starts building things. If you’re autistic and you know what hyperfocus feels like, you know what happened next. I didn’t answer the question normally. I mean, I did – “it means one out of two equal parts” – but by the time we pulled into the driveway, I was already designing a pizza game in my head.
Two hours later, pizzafractions.com existed.
How?
We sat down with Claude and laid out what we wanted. I knew what we needed from a technical perspective – TypeScript, Vite, mobile-first, accessible, GitHub for the code, Cloudflare Pages for hosting. He knew what we needed from a pizza perspective – and frankly, his credentials there are stronger than mine. A pizza gets sliced up on screen, some slices are there, some are missing, and you have to figure out the fraction. From there we went back and forth – “can we make the pizza look more like a real pizza?” and it would. “What if you could tap the slices to build your own fraction?” and it would. Over and over, fast enough that the idea never lost momentum.
It’s not magic, though. Knowing what “good” looks like matters. Early versions had real problems – answer choices that contradicted each other, fractions that didn’t match the pizza on screen, difficulty jumps that made no sense. You still have to stop Claude, redirect it, catch the stuff it gets wrong. Years of building software is what let me steer the conversation instead of just accepting whatever came back.
And honestly, my best design partner was my kid. Involving him in the process – “does this make sense?”, “is this too hard?”, “what would make this more fun?” – shaped the game as much as anything I told Claude. He found things I never would have – confusing wording, buttons that were too small to tap, levels that felt unfair. Claude is genuinely productivity-enhancing – but it’s enhancing your productivity, not replacing it.
The game draws actual pizzas using SVG – crust, sauce, cheese, pepperoni, the whole deal. I have never in my life spent this much time thinking about fake pepperoni placement. When you get an answer right, it plays a little chime and the pizza does a happy wiggle. Get three in a row and confetti rains down. Get five and it plays a triumphant little fanfare. These are not details I would have prioritized if I was grinding through this over a weekend. But because the boring parts were fast, I got to spend my time on the fun parts – the stuff that makes a kid want to keep playing.
My kids didn’t notice they were learning. They thought they were playing a pizza game. That’s the whole point.
The Speed Is the Point
My kid asked a question, and instead of pulling up a worksheet or downloading some ad-riddled app with a monthly subscription, I built something for them specifically in a couple hours. Without Claude this is a weekend project. By the weekend, the moment would have passed. Kid asks a question in the car, kid is playing a custom-built learning game the same afternoon. And now it exists – online, free, for any kid to use. That’s new.
If you’re a parent, it’s worth considering what AI can do for your kids’ education. Not as a replacement for teachers or classrooms – as another tool in the bag. A way to meet your kid where they are, when they’re curious, on their terms. And teach them how to talk to AI themselves – it’s a skill they’re going to need.
Jensen Huang, NVIDIA’s CEO, has been saying for a while now that learning to communicate with AI is more valuable than learning to code – that there’s a new programming language called “human.” I tend to agree. The earlier kids start learning how to ask good questions, the better off they’ll be.
If you’ve got a kid learning fractions and want to give it a shot, it’s free, there are no ads, and there’s no tracking: pizzafractions.com.