Cet article n'a pas encore été traduit dans ta langue — tu lis l'original en anglais.
How we built Spirkz
The short-video feed is the most effective attention machine ever built. We decided to point it at something worth keeping.
Somewhere around eleven at night, half the people we know are in bed watching strangers dance, cook, or fall off things. Nobody plans to. The feed simply has no bottom, and the swipe costs nothing. An hour later the light goes off and nothing remains from any of it.
Our starting point was that the format is not the villain here. Vertical video, thirty seconds, the next one already loading — that mechanism accidentally solved the hardest problem in education, which was never explaining things well. It was getting people to show up. Teachers have known forever that attention comes first; the short-video feed captures more of it than any classroom in history. The only thing wrong is the payload.
So the idea behind Spirkz is almost embarrassingly simple: keep the mechanism, swap the payload. A feed that behaves exactly like the one you can't put down, except every clip is a lesson — a physics idea, a story from history, a code trick, three minutes with someone who actually knows what they're talking about.
Instead of doomscrolling, you go spirkzscrolling.
One idea per clip
Thirty seconds is not enough to teach a subject. It is exactly enough for one idea — why the sky is blue, what a closure is, how the Ottomans took Constantinople. That constraint became our editorial rule: every clip carries a single idea, stated plainly, with no intro and no filler. The clip plants the hook; curiosity does the rest. Watch twenty of them in a row and you haven't studied anything, but you've collected twenty threads worth pulling, and that turns out to be how a lot of real learning starts.
Why the sky is blue, in one sketch
Closures, explained with a backpack
The chain that guarded Constantinople
Where "salary" actually comes from
A boring stack, on purpose
We are a tiny team, so every technical decision optimizes for one thing: shipping. The app is React Native with Expo — one codebase for both stores, with EAS doing the builds and the over-the-air updates. The API is Node.js. Authentication runs on better-auth, push notifications ride Firebase, and production lives on a plain Ubuntu server we manage ourselves. None of this will impress a conference audience, and that is the point. The interesting problems in Spirkz are the feed and the content, not the plumbing, so the plumbing had to be the most predictable thing we could find.
Shipping was still the hard part. The build pipeline fought us, the credentials fought us, and store review is its own kind of education. Google's sign-off came in July, Spirkz went live on Google Play in August, and the iOS build is sitting in App Review as this is being written. The first invitations went out to the waiting list this summer, and the first place it caught on is exactly where we built it: Malta.
What's next: the brain
The best educational material already exists — as long videos. So the next thing under construction is what we call the brain: a service that takes a five-minute tutorial and cuts it into feed-ready shorts along the natural boundaries of the spoken sentences, so each cut lands where an idea ends rather than mid-word. A creator uploads once and gets back a set of clips that already follow the one-idea rule. That closes the loop — the feed teaches, and the brain keeps it fed.
Spirkz is on Google Play today, with iOS close behind. A feed is going to eat an hour of your evening either way. It might as well leave something behind.