Up, But Not Tryable — A Report from the Reject Pile

One's Vibe lists AI-built products you can actually try. The bar for getting in is set deliberately wide — and even so, more than a quarter of what reaches the review queue never makes it to the gallery. Here is an honest look at the 615 projects we turned away — and the one pattern that dwarfs all the others.

Scope, stated up front. Everything below is the review stage only. By the time a project is counted here it has already cleared discovery's coarse filters: it's an AI-built product, it isn't a duplicate, and its URL answered a first ping. So "we published 71%" means 71% of what reached review — not of the raw firehose, which is far larger and filtered earlier. We're measuring the last gate, not the whole road. And most of these verdicts are automated; no editor agonized over 615 of them.

The shape of the funnel

Of 2,281 projects that reached review, 1,617 cleared it and were published (71%), 615 were rejected (27%), and 49 were archived, buried, or still in flight. Ten of those published projects are currently hidden and not shown, so 1,610 are live in the gallery as you read this.

Reachable isn't tryable

You would expect the top reason for rejection to be "dead link." It isn't — the checker weeds those out before review. The dominant reason, by a wide margin, is that the page is up and you still can't try it:

But "not tryable" holds two very different things, and telling them apart is the point.

Roughly two in five aren't a failing at all. Read from the review notes, about 40% of these are real products that simply aren't web apps — a native macOS or Windows app you download, an iOS or Android app from the store, a CLI, an SDK, a browser extension. The website is only an introduction; the product lives behind an install. Many are excellent. They are turned away for exactly one reason: One's Vibe lists web apps you can try in a browser, and a native or mobile app can't be tried online. That's scope, not quality — the most important sentence in this report. If we rejected yours on these grounds, it isn't a verdict on your work; it's that your work lives somewhere we don't index.

The rest are genuine shortfalls. The login wall with no demo. The waitlist that hasn't opened. The marketing page whose only button is Book a call. The screenshot that turns out to be a mockup. The page that never finished building. Here the medium is right — it is a web page — but there's nothing yet to try.

Every one of these, native app and gated page alike, returns a clean 200.

Liveness — the thing our public "Live checked" badge asserts — is necessary but not sufficient. A URL responding proves the lights are on; it doesn't prove there's a door you can walk through. 84% of what we reject fails that second, harder test. That gap is the whole reason One's Vibe exists: seen isn't tried is our north star on the demand side, and reachable isn't tryable turns out to be its mirror on the supply side.

Who clears the bar, by where they came from

<small>*small samples — read as directional.</small>

Two things surprised us. First, survival is remarkably flat across the social sources — around 70% whether a project was launched on X, posted to Show HN, or shared on Reddit. We expected Show HN, with its engineer audience, to clear the bar more often; it clears slightly less than X (67% vs 71%). A launch is a launch; the tryability bar doesn't much care which crowd witnessed it.

Second, GitHub is the outlier at 47% — unsurprising once said out loud: a repository is code, not a product you can try, and about half never resolve to something a visitor can click and use.

The 2% we won't describe loosely

Fourteen projects were rejected as harmful — scams, credential traps, the occasional thing pretending to be something it isn't. It's a small number and we won't dress it up. One's Vibe does not certify safety, and being published carries no guarantee. But "we don't certify safety" has never meant "we pass obvious harm through." Fourteen is the count of the second kind. We don't name them, and we never will — a browsable list of things we flagged is the one page we refuse to build.

What the reject pile is, and isn't

It isn't a graveyard. The Graveyard is for projects that were live and then died — a record of what people finished. This is the other honesty: the record of what asked to come in and didn't clear the door. Most of it never became a page you'd want to see, which is exactly why you don't.

If we turned yours away and you think we got it wrong, the door isn't locked. Fix the thing we flagged — open the demo, drop the login wall, ship past the waitlist — and ask again.