From swipe to algorithm: why AI is redefining online dating (and what it can't replace)

By Zendoric · June 28, 2026.

There is one figure that says it all: Bumble lost 21% of its paying users in the first quarter of 2026 compared with the previous year, falling to 3.2 million. Tinder, with some 50 million monthly users, saw its active base drop 7% year-over-year in March. The dating model based on swiping —which revolutionized digital romance more than a decade ago— is showing structural cracks. Not because people have stopped looking for a partner, but because the process has become exhausting and, in the opinion of many users, needlessly inefficient.

It is into that void that artificial intelligence is trying to push hard. The most striking case among those covered by the LA Times is Known, a startup founded in 2025 and based in San Francisco, whose CEO Celeste Amadon —who dropped out of Stanford to create it— argues that the big platforms have for two decades been 'designed, tuned and redesigned not to work.' Known's proposition is different from the very first contact: the user does not see profiles, does not swipe through photos; instead, they hold a fifteen-minute conversation with an AI voice that probes their personality, romantic history, lifestyle and expectations. The system then generates a match accompanied by a written compatibility summary. The startup, which has raised close to 10 million dollars from investors such as Forerunner Ventures and NFX, charges for each arranged date —15 dollars— on the argument that this guarantees users actually show up. It launched the app in February 2026 and plans to expand to San Diego in July.

In the segment of established apps, the responses are equally telling. Bumble has announced that it will eliminate the swipe in certain markets before the end of 2026. Grindr —which already operated with a grid of nearby profiles, without swiping— is testing a subscription tier called 'Edge' in Australia, New Zealand, the United States and Canada that includes AI tools to summarize relevant conversations, show personalized recommendations and predict the likelihood of a match. Its Chief Product Officer, AJ Balance, notes that some users are willing to pay up to 350 dollars a month because they perceive tangible value in the time savings. It is a figure worth attention: we are looking at a willingness to pay for romantic productivity that just a few years ago would have seemed implausible.

At the opposite end of the price spectrum, Facebook Dating offers its AI assistant for free and already has more than 21.5 million daily users worldwide. About one million people in the United States and Canada use the assistant each day to filter profiles by specific preferences. The product manager for this feature, Neha Kumar, neatly sums up the problem AI aims to solve: 'It's very hard to understand and find someone compatible with you based solely on your preferences by browsing profiles.' And from the founders' side, Justin McLeod —co-founder of Hinge— is working on a new app called Overtone on the premise that AI can create 'a totally new way' of finding a partner, more personal and more efficient.

The underlying logic is sound. If language models are capable of inferring compatibility from conversation patterns, expressed values and prior behaviors, they can do in seconds what a human would take weeks to discover through superficial message exchanges. Efficiency is no minor value when we are talking about one of the most decisive decisions in a person's life.

But there is a tension that no algorithm has yet resolved, and which the LA Times article captures precisely through the testimony of Marie Lansley, a Known user: 'Chemistry is always going to be analog.' Lansley took part in the AI matchmaking process, received a match with a compatibility summary and, even so, was not interested in the person. AI can optimize pre-selection, but it cannot guarantee the spark that happens —or does not— when two people sit face to face. And there is something more troubling than the mere lack of chemistry: the same technology that promises better matches also makes it easier to build inflated profiles with manipulated photos and messages written by chatbots, creating a gap between the digital person and the real one that can be disappointing at best and deceptive at worst.

As sector context, the shift toward the in-person is significant: users who abandon the apps are not giving up on meeting people, but are signing up for running clubs, pickleball classes or speed dating events. Known, in fact, organizes parties and in-person meetups in San Francisco to attract users. The paradox is eloquent: the most technological app on the current scene uses physical events as its acquisition engine. AI is not replacing human socialization; in the best scenario, it is preparing it better.

The real challenge for this sector in the coming years will not be technical but one of experience design: how to build systems that harness AI's inference capability without reducing people to compatibility vectors. Known's voice interview, Grindr's grid with scoring or Facebook Dating's conversational assistant are different approaches to the same problem. None has yet demonstrated at scale that it significantly improves outcomes in terms of lasting relationships —Known has not even published user figures— but the direction is clear: the swipe as the dominant interface has its days numbered. What comes next is, literally, still to be built.