Coffee Meets Bagel
The dating app that thinks the problem isn't supply. It's noise. Three sisters, one bet, and a curated list of people you might actually like.
CMB - the bagel that picks you back
The dating app that thinks the problem isn't supply. It's noise. Three sisters, one bet, and a curated list of people you might actually like.
It is 7:00 a.m. somewhere. A phone buzzes. Not with a hundred new matches. With one. Maybe a small handful, on a good day. The user taps. Reads. Decides.
This is what dating looks like inside Coffee Meets Bagel - a deliberately small portion in an industry built on portion-blindness. Where Tinder hands you a buffet and Hinge hands you a flight, CMB hands you a saucer. The pitch is almost rude in its modesty: maybe you don't need more people. Maybe you need fewer, better ones.
It is an unfashionable position. The unfashionable positions are often the ones worth holding.
Around 2011 a strange thing happened to dating. It became, by accident, a slot machine. The interface optimized for engagement. Engagement, of course, is the opposite of resolution. An app that helps you find a partner has, by definition, lost a user. So the apps quietly stopped trying.
The sisters Kang - Arum, Dawoon, and Soo - noticed. They noticed in the way only people who are not the target audience can. Most dating products, they observed, were built by men for men: an arcade of faces, no consequences, very few replies for the women on the other end. The numbers were ugly. The experience was uglier.
So they asked a small question with a big answer: what if a dating app behaved less like a casino and more like a thoughtful friend?
^ The thoughtful-friend hypothesis. Cheaper than couples therapy. Less reliable than a grandmother.
Arum had the idea in 2011. The twins - Dawoon, who'd been at Harvard Business School, and Soo, the eldest, a designer - did the unreasonable thing siblings do when one of them is on to something. They quit their stable jobs and joined her. The app launched April 17, 2012, in New York.
The product was opinionated from day one. Matches were called Bagels. Currency was Beans. You got a small daily batch - not a feed. If you didn't act, the match expired. Inconvenient. Also: clarifying.
By 2015 they were on stage at Shark Tank, asking for $500,000 for 5%. Mark Cuban, in a moment that has since entered start-up folklore, offered to buy the entire company for $30 million in cash. The largest offer in the show's history at that point. The sisters smiled politely and said no.
The internet, predictably, lost its mind. The Kangs received emails calling them, in roughly equal measure, crazy, greedy, and stupid. They kept building.
^ A decade of compounding small bets, two cyber-headaches, and exactly one Mark Cuban.
The CMB app is, in the end, a piece of software with a strong opinion about your time. It uses behavioral signals, declared preferences, and a matching algorithm to deliver a hand-picked daily set. There is no infinite feed. There is a cap. The cap is the point.
Conversations open with ice-breakers. The discovery tab lets women, in particular, make the first move - a feature that predated several louder competitors. Beans buy you extensions, do-overs, and access to people slightly outside your usual perimeter. Premium subscriptions unlock read receipts and additional filters.
None of it is revolutionary in 2026. All of it, in aggregate, is unusual. Most dating apps treat their users like data points. CMB treats its users like guests at a small dinner party who, frankly, would prefer not to talk to most of the room.
^ Yes, the matches are actually called Bagels. Yes, the founders are aware that is a lot.
Skeptical readers - and these are the only kind worth writing for - want numbers. The numbers, gathered from public sources and the company's own disclosures, look like this:
^ Bars are illustrative, not arithmetic. They are scaled to vibes.
Help people fall in love and build real relationships - using fewer matches, better data, and an interface that respects your evening.
The dating-app industrial complex is finally cracking. Engagement metrics have peaked. Users are tired. Regulators are sniffing. A generation that grew up on the swipe is openly nostalgic for things that never existed - the mythical good first date, the unscheduled meet-cute, the friend-of-a-friend introduction.
CMB has spent fourteen years building software that resembles that last category. Not because it is romantic. Because it works. The model survives because it is closer to how people actually want to fall in love than the alternatives, and because every year a few more singles get tired of treating their romantic life like a Spotify queue.
There are risks. The dating-app market is brutal. A single bad press cycle or another data incident could undo years of trust. And the company is small relative to the public-market giants it spars with. But there is something deeply useful about a business that has, at multiple inflection points, said no - no to Cuban, no to the swipe, no to a feed that never ends.
Back to the morning. The phone buzzes. One Bagel. Maybe two. The user taps. Reads. Decides. The app, having done its job, gets quietly out of the way.