I’m Building an AI-First Dating App — Here’s the Hypothesis
Most dating apps optimize for engagement.
More swipes.
More matches.
More notifications.
But more doesn’t mean better.
I’m building Hullo around a different hypothesis:
AI should optimize for compatibility, not activity.
Instead of endless browsing, we’re experimenting with an AI Match button — a system that analyzes user intent, profile signals, and behavioral patterns to suggest higher-quality matches.
Some early lessons so far:
People say they want meaningful connections, but behavior often signals curiosity and entertainment.
Profiles are low-signal by default — we’re testing AI-assisted profile optimization to improve match quality.
The hardest part isn’t building the model — it’s aligning incentives so the product rewards better outcomes, not more usage.
We’re still early. Still testing. Still iterating.
Curious to hear from other founders:
If you were rebuilding dating from scratch in 2026, what would you optimize for — engagement or compatibility?
https://www.nytimes.com/2015/02/13/style/the-36-questions-on-the-way-to-love.html
Lot of research when into this. Could it be codified?
Any alternate sources of data that you are using for the profile optimisation or to optimise for compatibility? If yes, could you share some of them?
This is a really interesting direction. Optimizing for compatibility signals instead of swipes feels like the right problem to solve. Curious to see how the AI matching performs in real use — would love to test and share feedback.
Some people love the browsing too.
Hey that’s near. My friends actually had similar pain points - saying modern dating apps only focus on looks and not on feels or compatibility. Interesting how do measure “compatibility” though?
I think I would try to optimize for compatibility with the user's intent. Ai is very good at finding patterns if you are clear on what your a re looking for.
This comment was deleted 6 months ago