← From the matchmaker's desk

What a modeled conversation can—and cannot—predict.

By Find Your Person · Published

Product-mechanics articles are checked against the current implementation and public policies.

Before a potential pair becomes an introduction, Find Your Person models a conversation between representations built from the information each member has provided.

The simulation follows several conversation phases and produces an evaluation that includes an overall score and an estimated second-date likelihood. That evaluation must clear a gate before the pair can move forward. It also contributes to the broader weighted match score.

It is one signal among several.

The simulation currently carries 28% of the baseline weighted score. Photo-preference fit, reciprocal discovery preferences, interview-derived traits, engagement, and confidence make up the rest. Both members’ directional scores must pass. The system is deliberately broader than a single AI-generated conversation.

It does not simulate two real people in full.

A model works from profile representations, not complete human beings. It cannot reproduce body language, timing, physical presence, attraction in context, or the countless surprises of an actual conversation. A high result means the available inputs look promising under the model. It does not mean a date will go well.

Why use it at all?

Written preferences can say whether two people want the same broad future. A modeled exchange can add a different kind of signal about topic overlap, pacing, and conversational fit. Used carefully, it can help rank potential introductions and generate a clearer explanation for why one is being shown. The member still decides whether to continue.

You've done enough scrolling.

Let your matchmaker take it from here.

By joining, you agree to our Terms and Privacy Policy.

Free to join · No credit card · Confirm your email to save your place