Meta quietly launches vibe-coded gaming app Pocket
Meta's latest attempt to colonize the creator economy didn't arrive with a keynote, a blog post, or even a press release. It slipped onto the App Store and Google Play on June 29 — Pocket, a "vibe-coded" gaming app born from the ashes of Gizmo, a startup Meta acqui-hired earlier this year. The silence is the story. When a company with Meta's reach launches a consumer product without fanfare, it's not confidence; it's damage control.
The term "vibe coding" has become Silicon Valley's favorite euphemism for "we hope the LLM figures it out." Pocket lets users type prompts like "make a tiny platformer where a cat collects sushi" and spits out playable HTML5 experiences called "gizmos." There's a feed, of course — everything is a feed now — where you can swipe through other people's generations. It's Roblox without the engine, Unity without the learning curve, and arguably without the soul.
The Gizmo ghost
Gizmo, the app Pocket effectively replaces, had 635,000 lifetime installs and a 98% positive sentiment score, per Appfigures. That's not nothing. It's a modest, engaged community that actually liked the thing. Meta's playbook here is familiar: acquire the team, strip the brand, repackage the tech inside a sterile new shell, and hope the algorithm does the rest. The Gizmo app still sits in the store, a digital ghost town Meta hasn't bothered to delist. That tells you everything about the respect for existing users.
Alessandro Paluzzi, the reverse engineer who first spotted Pocket, has made a career of catching Meta's quiet bets. Business Insider and Investing.com amplified his find. Meta, asked for comment, offered silence. That's not oversight; it's strategy. A quiet launch lets you kill the product in six months without explaining why.
AI creation as content farm
Pocket is the logical endpoint of Meta's AI creator push: Meta AI for images, Vibes for video, Edits for… editing, now Pocket for "games." Each tool lowers the floor until there's no floor at all — just a prompt box and a feed. The promise is democratization. The reality is a content farm where the farmers are LLMs and the crop is infinite, forgettable sludge.
We've seen this movie. Meta's Horizon Worlds bet on user-generated VR spaces; it's a wasteland. Instagram Reels copied TikTok's algorithm but not its culture. Now Pocket copies Roblox's model but replaces the builders with a text box. The pattern is clear: Meta doesn't want to build platforms. It wants to build feed stock.
The "experiment" trap
Reporters will call Pocket an "experiment." That's the get-out-of-jail-free card. If it grows, Meta claims vision. If it dies, it was just a test. But experiments have hypotheses. What's Pocket's? That people want to play AI-generated micro-games between doom-scrolling Reels? That "vibe coding" produces anything worth keeping? The hypothesis is unspoken because it's uncomfortable: Meta needs AI to generate engagement inventory it doesn't have to pay creators for.
Gizmo's 98% positive rating suggests a small group found genuine delight in the tool. Pocket's feed will dilute that delight into algorithmic noise. The creator who spent hours prompting a clever physics puzzle sits beside the user who typed "funny game" and got a broken cookie clicker. The feed doesn't distinguish; it optimizes for retention.
A pattern of abandonment
Remember Lasso? Collab? Tuned? Hobbi? Meta's graveyard of quiet launches is deeper than Pocket's gizmo feed. Each arrived without announcement, each vanished without eulogy. Pocket joins a lineage of products Meta treats as disposable feature probes rather than platform commitments. The difference: Pocket carries the DNA of a team that built something people liked. Meta bought that team to strip-mine their insight for the main apps — Instagram, WhatsApp, the feed.
There's a version of Pocket that matters: a genuine creative tool with community features, curation, export, monetization — a home for the weird, specific joy of making tiny interactive things. That version requires investment, patience, and a company that values creators over content. Meta has repeatedly proven it's not that company.
The real product is the prompt
So why launch at all? Because every prompt typed into Pocket is training data. Every gizmo generated, played, swiped past, or shared is a labeled example of human intent meeting model output. Meta's Llama models need this signal. The app is a data harvester wearing a party hat.
Users sensing this won't stay. The 98% sentiment will curdle. The feed will fill with slop. And in twelve months, Pocket will join Lasso in the graveyard, unannounced, unmourned. The Gizmo team will be reassigned to optimize Reels ranking. The cycle completes.
Vibe coding isn't a platform. It's a parlor trick. Meta knows this. That's why they launched it quietly — they don't believe in it enough to shout, but they need it enough to try.