Coding with kids: how our family builds real apps together
We are a dad and two kids — Lily (10) and Matthew (8) — who build web apps together on weeknights. Everything below is what actually worked for us, including the parts we got wrong first. It is not a curriculum. It is closer to a family habit.
Our four steps
1. Start with their problem, not your idea
The best projects come from something a kid already cares about. Matthew was frustrated that he could never finish his Rubik's cube, so the app had an obvious job: help him solve it and teach him why each move works. If you have to explain why the idea is cool, it's your idea, not theirs.
2. Give each kid a real role
Our 10 year old runs the look and the words — colors, layout, button copy, what feels friendly. Our 8 year old is the tester and rule-maker: he decides what the app should do when you get it wrong. The parent drives the keyboard and the tricky parts. Nobody is just watching.
3. Ship something small the same day
A page with one working button beats a plan for ten features. We aim to get something on a real link in the first sitting so they can send it to a grandparent that night. That first share is what makes them want to come back to it.
4. Let feedback be the next sprint
When a friend says the buttons are too small or the story got boring, we write it down and fix it together next session. Kids learn more from one round of real feedback than from any tutorial.
Six house rules that keep it fun
- Sessions end on a win, not a bug
- We stop while something works. Ending on a broken screen kills momentum faster than anything else.
- 45 minutes is plenty
- Short and repeatable beats a marathon weekend. We usually do one or two sittings a week.
- They name everything
- App names, character names, button labels. Ownership is the whole point, and their names are better anyway.
- No accounts, no data collection
- For kid-made apps we keep things public, simple, and login-free. Less to explain, less to protect.
- Type it, don't dictate it
- Even when it's slower, the kid types the words that show up on screen. Seeing their sentence appear in the app is the moment it clicks.
- Publish, even if it's rough
- A real URL turns a craft project into a thing that exists. Polish later.
A real example: Matthew's Cube Buddy
Matthew could scramble a Rubik's cube beautifully and solve exactly one layer. The ask was simple in his words: "tell me what to do next." So we scoped it to that — enter what your cube looks like, get the next moves, and get them explained in kid language rather than cuber notation.
He chose the colors and decided the app should never say "wrong." Lily read the instructions out loud and cut anything that sounded like homework. The first version did one thing and did it on a live link the same week; everything after that came from watching his friends use it.
If you want to try this weekend
- 1Ask your kid what annoys them, or what they wish existed. Write down three answers, no filtering.
- 2Pick the one that can be a single screen. One input, one output.
- 3Sit them next to you and let them decide the title, the colors, and the first sentence on the page.
- 4Build only that screen. Stop when it works.
- 5Publish it and let them text the link to someone who will be excited.
- 6Ask that person for one thing to improve, and make that your next session.
That's the whole method. The code matters less than the fact that your kid made a thing, on the internet, that someone else used.