Meta’s Pocket and the Vibe-Coding Bet
Meta quietly launched a new app last week. It’s called Pocket, and it lets anyone type a prompt and instantly get back a playable mini-game — a “gizmo” — that responds to touch, tilt, your camera, even the music on your phone. Scroll a feed of other people’s creations. Play, remix, repeat.
TechCrunch broke the story on July 2. Business Insider followed up. The app is real, it’s live on both app stores, and Meta hasn’t officially announced it. That’s how you know they’re testing something they’re not sure about yet.
Pocket isn’t an original idea. It’s a near-exact copy of Gizmo, an app by Atma Sciences Inc. that had over 635,000 installs and a 4.9-star rating. Meta quietly acquired the Gizmo team earlier this year. Then they built Pocket. The clone is a feature, not a bug — it’s Meta’s playbook since Stories.
The counterargument is obvious and worth stating: acquisition and distribution is how good ideas reach scale. Gizmo was a niche app with great ratings and modest reach. Meta has billions of users, a proven ad network, and the engineering muscle to make something like this work at a global level. If you believe vibe-coding mini-games is a genuinely good idea — and I think there’s a real argument it might be — then you should want the team with the best distribution to run with it.
But here’s where it gets uncomfortable. Meta’s track record with creative tools isn’t great. They bought the Gizmo team, they’re cloning the concept, and the original app still exists — for now. The pattern is familiar: acquire, absorb, deprecate the original, then optimize the clone for whatever maximizes engagement metrics. The question isn’t whether Pocket will work. It’s whether it will work for the people making things or for Meta’s ad business.
The deeper story here is about vibe-coding itself. The term is stupid (it sounds like a blockchain startup from 2021), but the concept is real. AI lets you describe what you want and get something approximating it. For games — simple interactive toys, not Skyrim — that actually works. The threshold between “idea” and “working prototype” has dropped from weeks to seconds.
Sekai, a competitor in this space, just raised $20 million in Series A. TikTok has experimented with mini-game feeds. The pattern is converging: social feeds + AI-generated interactive content. It makes sense. Short-form video proved people will consume endless user-generated content. Short-form games are the next logical step — they’re interactive, they’re shareable, and every one is a conversation starter.
The skepticism I can’t shake: most vibe-coded games are going to be terrible. Shallow, samey, one-interaction toys that you play for eight seconds and forget. The ones that are good — the rare gizmo that actually surprises you — will get buried under an avalanche of slop. That’s the same problem TikTok solved with algorithmic curation, but TikTok has years of engagement data. Pocket starts from zero.
There’s a version of this future I like. A world where anyone can describe a game and have it exist, where creativity is gated by imagination instead of technical skill. The kid who can’t code but has a great idea for a game mechanic finally has a way to prototype it. That’s genuinely powerful.
There’s another version — the one Meta has earned the benefit of the doubt on — where Pocket becomes another feed optimized for time-spent, where “making” is something you do once before you go back to scrolling, and the gizmos are just ads wearing a Halloween costume.
I want to believe in the first version. But Meta’s history is a heavy thing to bet against.
Sources: TechCrunch — Meta quietly launches vibe-coded gaming app Pocket, Business Insider — Meet Meta’s New Vibe-Coding App: Pocket, Slashdot — Meta is Quietly Launching Pocket, seen via The Brutalist Report.