Love by algorithm: how AI agents are replacing swipes in dating

Online dating has existed for nearly thirty years, and throughout this time its primary mechanism has remained the same reflexive action: swipe, evaluate, swipe left or right. Tinder turned the search for a partner into an arcade game, Bumble added a feminist twist, Hinge promised "an app designed to be deleted." Now startup Fate claims it will end the very idea of the swipe — and hand the decision about compatibility to agentic AI. The technology industry has long sought ways to optimize human relationships. It seems it has finally decided it has found one.

The principle behind Fate differs drastically from what millions of dating app users have grown accustomed to. Instead of an endless feed of photos, the app launches an AI assistant that conducts an extended interview with the user. The system asks questions about values, life goals, habits and communication style, analyzes speech patterns, and based on the collected data, generates a list of five potential partners. No lottery, no chance — only an algorithm that, according to the developers, knows more about you than you yourself managed to understand in the first minutes of conversation.

The logic of the app's creators is understandable and in some sense convincing. "Dating fatigue" — a term psychologists use quite seriously — has become a mass phenomenon. Hours spent on meaningless conversations, disappointment from expectations not matching reality, the feeling that searching for a partner has turned into an exhausting side gig — all of this created demand for something fundamentally different. Fate promises to shorten this path: instead of hundreds of viewed profiles — five precise matches. This sounds like a solution to a real problem, and that is precisely why the project has already attracted the attention of both investors and media.

However, behind the elegance of this construction lies a question that the developers have yet to answer clearly: what exactly happens when an algorithm takes on that part of human experience which is inherently unpredictable? Romantic attraction is one of the least formalizable phenomena. People fall in love against common sense, against "value compatibility" and against any speech patterns. History knows countless couples who by all rational criteria should not have been together — and it was precisely they who proved to be the most durable. An algorithm optimized for "observable complementarity of linguistic patterns" risks systematically filtering out exactly such possibilities.

Fate's critics also point to a broader problem: the app does not simply help the user make a choice, it makes that choice for them. This is a qualitative shift compared to ordinary recommendation algorithms. When Spotify selects a playlist or Netflix recommends a series, the cost of error is low. When agentic AI decides who you should meet with the expectation of long-term relationships, the stakes are entirely different. The very architecture of the app assumes that the user trusts the system to interpret their own desires — and this trust can become a trap. A person accustomed to delegating such decisions to an algorithm gradually loses the skill of independently figuring out what they want.

Fate, however, reflects a much broader trend in the technology industry. Agentic AI — systems capable not just of answering questions but of independently executing multi-step tasks and making decisions in the user's interest — is rapidly moving beyond corporate tools. After agents mastered travel planning, email management and online shopping, the transition to more subtle social spheres was only a matter of time. Dating turned out to be the next frontier — and this speaks volumes about how far the industry is willing to go in automating human life.

Ultimately, Fate poses a question before us that is not technological but philosophical. Are we willing to accept that the most important decisions in our lives can be the result of an optimization task? An interview with an AI assistant is still not a conversation with another person, but a simulation of such a conversation, aimed at extracting data. If the search for a partner ultimately becomes an algorithmic transaction, we will gain in efficiency — but risk losing something that we do not yet know how to measure and enter into a database.