Even Bumble Admits the Swipe Is Broken

On May 7th, Bumble CEO Whitney Wolfe Herd told investors that the company is going to retire the swipe — the gesture that built her company, the gesture that defined an entire decade of online dating, the gesture that her users now describe as “exhausting” and “fatiguing.”^[2] She said the swipe has “degraded their love lives.”^[2]

That is a remarkable thing for the CEO of a swipe-based dating app to say. It is more remarkable when you look at her stock price, which has lost roughly 90% of its value since 2021.^[5]

The official framing is innovation: Bumble is moving past a tired UI pattern and toward something newer, smarter, AI-assisted. Wolfe Herd called it “revolutionary for the category.”^[1] The unofficial framing — the one her balance sheet tells — is simpler. The product never worked the way the marketing said it did. After twelve years of swipe-driven engagement, the market is finally pricing in the difference.

What the numbers actually say

Bumble's paid users dropped about 21% year-over-year, from 4 million to 3.2 million.^[1] Match Group, the parent company of Tinder and Hinge, reported its 31st consecutive month of Tinder monthly-active-user decline on its May 5th earnings call.^[3] Tinder paying subscribers fell 5% to 8.6 million.^[3]

The Gen Z number is the one worth sitting with. A 2024 Forbes Health survey^[5] found 79% of Gen Z users report fatigue from dating apps. That cohort is now in its mid-twenties and supposed to be the engine of the next decade of dating-app revenue. Instead, they are the cohort writing the eulogies.

Dating apps have a retention problem dressed up as an engagement success.

Here is the trick: a swipe is engagement. A swipe is a measurable, ad- sellable, dashboard-friendly micro-event. But a swipe is not a date, it is not a conversation, and it is increasingly not even a flirtation. It is, structurally, a vote on a stranger's headshot — and after a few hundred of those, the human brain stops treating any of them as real people. This is what users mean when they say the apps are exhausting. The fatigue is not from too many matches. The fatigue is from too many micro-decisions about images of people you will never meet.

Why the AI rescue plan won't fix it

Every major app has reached the same conclusion: the swipe is finished. Tinder, in February, announced an AI feature called Chemistry that lets users “get just a single drop or two, rather than swiping through many, many profiles.”^[4] The feature scans users' camera rolls to extract personality signals.^[4] Bumble is building an AI assistant called Bee.^[1] Hinge, which is still growing — 28% year-over-year revenue, 15% paying user growth^[3] — has its own “AI Matches” feature pulling from short prompts.

Notice the common pattern: each company is grafting AI onto the same product they had before. The input is still photos. The unit of interaction is still a binary judgment. The data the AI has to work with is, almost always, what your face looks like and what gym you go to.

This is the wrong layer to put intelligence on. You cannot extract a person's emotional vocabulary from a beach photo. You cannot read attachment style off a sunset on the rim of Half Dome. You cannot find out how someone handles silence, or grief, or a Tuesday night, by looking at the photos they took of themselves. The data was never there.

What a post-swipe app looks like

We built AISURU on the premise that the input to a dating app should be what people actually think, not what they look like in good lighting. So AISURU asks you to write five short essays — about 300 words each, roughly twenty minutes total — about how you spend a Tuesday, what you believe against the grain of your peer group, how you've loved and been loved, what you're ambitious about, and what you've recently changed your mind on.

AI then reads those essays the way a thoughtful matchmaker would. It pulls personality traits across four weighted dimensions — lifestyle (35%), emotional depth (30%), complementary differences (20%), and values (15%). Then it scores compatibility against everyone in your city who's also written essays. Only pairs scoring above 65 out of 100 surface as matches. The system runs once a day, overnight. You wake up to people you would actually want to spend an evening with.

There is no feed. There is no swipe. There is no infinite scroll. The product was deliberately built to not be sticky in the time-on-app sense, because the goal was never to keep you on the app — the goal was to get you off it, into a coffee shop, with a person whose mind you already know a little.

The thing that's actually shifting

The honest read on Bumble's announcement is not that swiping is over. It is that the swipe was a billion-dollar UI hack that worked once and stopped working when the users got tired of being treated as decks of cards. The next decade of dating products will sort along two axes. What they take as input — photos, swipes, prompts, essays, voice, camera roll. And what they optimize for: time-on-app, or time-off-app-and-with-a-person.

The first category is the one Bumble and Tinder are still in. The second is the one we're building.

Time-on-app and time-with-a-person are not the same metric. After a decade of pretending they were, the market is finally noticing.

A note on the AI we use

Because every dating app is announcing AI features this season, it is worth being specific about what AI does and does not do on AISURU. AI reads the essays you wrote, extracts personality signals from them, and scores compatibility. AI does not generate profiles. AI does not impersonate users. AI does not chat on anyone's behalf. Every essay on the platform was typed by the person you're reading; every conversation is between two humans. We documented the full scope on our Trust & Verification page.

If that distinction sounds boring, it is, by design. The interesting thing on a dating app should be the other person, not the technology between you.