Last November, my 9-year-old asked me if we could "make a game about the thing we learned in school." That week the topic was housing around the world. She wanted a game where you match the family to the house.
We sat on the sofa with an iPad. I opened Google AI Studio. She talked. I typed the parts she could not type yet. Her younger sister, 7, climbed over the back of the sofa to watch. An hour later, there was a game on the screen with her characters, her rules, her world. She ran to show her mom, shouting, "Look, the game I did!"
That Friday became a ritual.
What they actually built
Over the last six months my daughters have built dozens of small games. A matching game about houses. A dress-up game for their favorite cartoon. A quiz about animals of the savanna. A birthday card for their grandmother, written in three languages. A memory game built around photos of our dog.
None of these are polished. Most are silly. Some are thirty seconds of fun and then forgotten. That is not the point.
The point is the loop. They have an idea. They say it out loud. Something real appears on the screen. They play it. They notice what is wrong. They say the next thing. The next version appears. They learn what a bug is, not from a textbook, but because the character won't sit on the chair. They learn what "clear" means, not from a writing class, but because the AI didn't understand "that thing, you know, the one."
Articulation became a skill. Iteration became a habit. Shipping became a normal afternoon activity.
Why I think this matters
My daughters will grow up in a world where, for most problems, the distance between "I have an idea" and "I have a working thing" is a well-formed sentence.
The kids who understand that early will treat ideas differently. They won't file them away under "one day." They'll try the small version tonight. If it works, they'll build on it. If it doesn't, they'll learn something and move on. That muscle, the one that converts imagination into reality without waiting for permission, is the entrepreneurial muscle. It used to take years to develop. Now you can start at 7.
What they need is not a coding curriculum. My daughters cannot code, cannot touch-type, cannot read English fluently. What they need is practice at the thing coding used to gate: turning a fuzzy idea into a clear instruction, and then caring enough to make the second version better than the first.
That is a mindset, not a syllabus. You teach it by doing it together on a Friday afternoon.
I wrote about the early version of this with my daughters in Inside AI Agents a few months back, if you want the longer story: How my daughters built their own games with AI coding agents.
The podcast moment
I assumed this was a me-and-my-kids thing.
Last month I was listening to a founder on a podcast. In the middle of talking about his company, he mentioned, almost in passing, that his son had been building games with Replit. He said it the way I talk about my own Friday afternoons. Not as a novelty. Not as a party story. As a normal thing his kid does.
That was the moment it clicked. There are already kids doing this, all over the world, on whichever tool their parent happens to have a tab open in. The generation that grew up voice-prompting tablets is about to start building with them. They will learn the skill one way or another.
The question is what they'll learn it on.
Why a product for kids, specifically
General-purpose AI tools were not built for 7-year-olds. They ask for logins, throw walls of settings, assume an adult is reading, and crash or refuse in ways that are not kid-appropriate. Every Friday my daughters need me in the loop, not because I'm adding value, but because the tool keeps tripping over things a child can't handle on their own.
Kids deserve their own thing. A product where the scaffolding matches how they actually think. Voice-first, because typing is still slow for them. Patient, because their ideas evolve mid-sentence. Visual, because a block of text does not feel like a game. Safe, because parents should be able to hand the iPad over without watching. Proud of them, because finishing something should feel like finishing something.
Not a stripped-down adult tool. A real creative environment with kids at the center.
That's what we are building.
Why now, and with whom
A few weeks after the podcast, I called Alex. Alex and I built a company together before, in another life, at Always. We've shipped things together, argued about things together, and know how the other one works when it's 11pm and the demo is tomorrow. I told him what I was seeing with my daughters, and the podcast moment, and what I thought the product should be.
He was in the next week.
We called it Buildaloo. Loo is an octopus, because an octopus has eight arms and so does a curious 7-year-old. The mascot is coral, the tone is warm, the goal is simple: a kid should be able to walk up, tell Loo an idea, and walk away with something real that they made.
We are at the beginning. If you are a parent, or someone who cares about what tools kids grow up holding, we'd love your eyes on what we ship next.
Joao & Alex