Tinder CTO Says Top Engineers Spend $15,000 a Month on A.I. Tokens
Modern dating is increasingly shaped by A.I. Top dating apps like Tinder, Hinge and Bumble are all experimenting with A.I.-driven features ranging from matching algorithms and A.I. assistants to conversation starters and profile photo selection. Behind the scenes, meanwhile, A.I. coding agents are also changing how the apps themselves are built.
Over the past year, Tinder, the world’s biggest dating app, has become increasingly A.I.-pilled. “We’re using A.I. in every stage of the dating journey,” Vinay Kuruvila, CTO of Tinder, tells Observer in a video interview. At its parent company, Match Group, he said A.I. costs rose from $1 million a year at the beginning of 2026 to $10 million a year.
While Kuruvila said the company wasn’t “tokenmaxxing,” its push to increase A.I. adoption has nonetheless driven a ten-fold increase in token spending from January through August, as it worked on “trying to get everyone in the company A.I.-pilled.”
Kuruvila said A.I. coding tool adoption among Tinder engineers has surged from around 10 percent when he joined the company to nearly 100 percent today, with A.I. now generating more than 90 percent of its code and significantly accelerating development. Tinder is now roughly three times faster at shipping code changes and feature improvements, he added.
Today, Tinder’s top engineers are spending around $15,000 a month on tokens. While recommendations is one of its most A.I.-pilled teams, the company’s number one user sits on the profile team. Kuruvila said Tinder’s most productive engineers are using autonomous agents that run 24/7.
Tinder uses A.I. across everything from recommendations to photo selection, though Kuruvila said the area where the company has seen the most success with A.I. and machine learning is its recommendation algorithms, which have changed significantly since he joined as CTO in August 2025.
With recommendations, the goal is to show users people “they are very likely to vibe with,” Kuruvila said, based on common interests and the likelihood that they will have good conversations and, ultimately, go on real-life dates. One of Tinder’s core internal metrics is Sparks, defined as conversations that have gone back and forth six times, which the company considers a strong indicator that a match could lead to a date.
A.I. is also helping Tinder bring new features to the app. Most notable is camera roll scan, an opt-in capability that Tinder plans to roll out globally “soon.” The feature can scan a user’s camera roll and select the strongest photos for a profile, alongside other profile-improvement tools that have yet to be announced.
Tinder is testing a feature called camera roll scan in several markets that, with users’ permission, analyzes their photos and selects the ones most likely to strengthen their profiles. The company has found that many users, particularly men, struggle to identify their best photos, and the feature is performing well in testing.
“If you put a shirtless gym selfie because you think that’s your best photo, but most women are turned off by it, you’re not going to get outcomes,” Kuruvila said. “People don’t know how to best represent themselves on dating apps. I think A.I. has tremendous potential to help someone represent themselves authentically and in a way that serves them well on their profile.”
There is also a major launch coming to the platform later this year that Kuruvila refers to as “real-time recommendations.” He compares it to TikTok, where liking one video can quickly prompt the algorithm to serve another on the same topic. Currently, Tinder can take up to four hours to detect changes in a user’s swiping behavior. By the end of the year, that process will happen in a matter of seconds.
Tinder’s A.I. transformation comes as Match Group invests more heavily in A.I. across its portfolio of dating platforms. CFO Steven Bailey told Observer in July that the company had given all 2,300 employees access to tools like Claude and said Hinge’s A.I.-driven matching system had increased matches by 15 percent since being rolled out a few quarters earlier.
The bigger question is whether more A.I. is actually what online daters want. A study commissioned by Match Group surveyed 1,000 people between the ages of 18 and 39 and found that 47 percent of singles had a negative view of A.I.’s use in romantic contexts. At the same time, commentators have raised broader concerns about whether A.I. can meaningfully encourage connection between users.
Whether Tinder’s bet on A.I. can help address the loneliness epidemic will ultimately depend on how effective its algorithms are at getting people off the app and onto actual dates. Few users are likely to complain about A.I. if it leads to better matches. But implemented poorly, the same technology designed to improve connection could end up overwhelming the very human interactions that dating apps are meant to facilitate.
