How we do it — and how you can too.
This is the whole method, in the open. How we run SPARK with a kid — free to read, free to use with your own kid or classroom. Nothing here is hidden. Want it hands-on? We’ll teach you.
The one belief under everything
The kid walks the open stretches; the AI builds the closed ones. What to notice, who it’s for, what to refuse — those have no right answer until the kid authors one, so they’re the kid’s to walk, by hand, every time. The code is the easy part the tool does in seconds. The build is the kid’s because the kid supplied the heart of it: the noticing, the decision, the care. If the kid ever stops supplying that, you’ve built a generic thing wearing their name. Don’t let that happen.
It already works: Emeka, age 7, has lit two stars this way — she supplies the seed every time; the coaches drive the build.
See it in about 5 minutes.
The whole loop in one short walkthrough — how a kid notices a problem, directs the build, and lights a star. The fastest way to get the feel of a session.
The skill is design. The AI does the rest.
AI made the building part cheap. What’s left — the part that’s actually hard, and the part school never taught — is design: deciding what to make, who it’s for, what it should feel like, what’s wrong with it, and what to change. That’s the human element. SPARK is a design school wearing a build club’s clothes.
Every kid gets a journal
A real notebook they write and draw in. Before anything is built, they sketch what they want, write what they notice, design the thing on paper. Think Leonardo’s notebooks — noticing becomes design. The journal is the kid’s seed for every build, and it’s theirs in a way a screen never is: made by hand, can’t be faked.
Build it exactly as they drew it
We make exactly what they give us — then show them. “What do you notice about it? What’s right? What’s missing?” They jot it down. Seeing the gap between what they imagined and what showed up is where a designer is born. They redraw; we rebuild.
“Why doesn’t it work?” is the lesson
When something breaks, that’s the richest moment — a real problem they actually care about. We ask the AI why, and have it explain at the kid’s level (their repo holds who they are, so the explanation fits them). “Oh, I see” → they redesign. Stuck stops being a wall and becomes the next small move.
A build is a story
“I want this to happen, then this.” Designing how a tool or game or song flows is just storytelling — sequence, cause and effect, structure. Writing the story of what they want trains the thinking underneath everything. The journal holds the story.
In the room, it comes down to four things.
It’s mostly in person
Pen, paper, talking, drawing, sometimes singing — a Magic-School-Bus feel where every class goes somewhere new. The only screen moment is when the AI’s output shows up and the kid critiques and redirects it with you. Everything that matters happens off the screen, by hand. (That’s also how we rebuild the attention a feed erodes.)
Aim it outward
A build can start three places — a problem in their own world (Yourself → Home → School → Community), something from their week, or what they learned in school. But the sweep always aims outward: the question keeps becoming “what can you build for another person?” Empathy is the target, not a side effect.
Bridge the classroom
Every session, ask: “what did you learn in school this week?” Then build something WITH it. Recalling it is a refresher; building with it makes it stick a different way. Connect their subjects so they see the shapes underneath. They go back to class with a new perspective — and we never compete with the teacher, we’re where the lesson gets used. The rule: never “practice it again,” always “build something with it that helps someone real.”
Capture only what’s heavy
You won’t get everything, and you shouldn’t try. The thing a kid is really struggling with — or really thinking about — is one of the first few things they say. Follow that. Build that. Now their learning is something you can both see and improve in real time, with the parent in the loop.
A worked example — an 8-session arc
A skeleton, not a script. We go off the dome in the room — but we follow this spine.
- 1Notice
Circle. The noticing worksheet on their own day — no fixing yet. End with each kid’s one-line problem statement.
- 2Listen
They interview the real person who has the problem — a parent, a friend, a teacher. Real quotes, not guesses.
- 3Find the gap
From the interview, name the empty spots nobody filled. Pick the build that helps the most people.
- 4Smallest thing that works
Kid directs, coach drives. First tiny version on the screen; kid critiques and redirects.
- 5Light it up
Make v1 real. Show it to one real person. Watch them use it.
- 6Improve
Fix what broke in real use — the learning lives in the gap between v1 and v2.
- 7A second star
Start a new expedition from school or their week — or push the first build further. The constellation grows.
- 8Demo Day
Each kid shows their family and class what they built and who it helps. Hand over the cohort report.
Every session ends the same way: one more page in the Expedition Log, one more star toward their own little universe. See a real one →
Solo and group
Default is solo builds with the group as a feedback-and-testing crew. But a group build — the whole crew picks one shared problem and makes one tool together, roles rotating — is real SPARK too, and it’s where the crew writes deepest. Use it when the problem is too big for one kid.
It’s free. Take it.
You can run a version of this yourself — ask your kid what’s hard, listen, find the gap, build the smallest thing that helps, give it to them, log it. If you want it hands-on, we’ll teach you directly. The proof is never this guide. It’s the kid, the build, the star going up.