The apps got the
order wrong.
Every app that introduces people starts with what you look like and works backwards towards who you are. Seam starts with who you are and trusts the rest to follow. This is an explanation of what we measure, how we match, and why we think the order is the whole ballgame.
The bet
Attraction is real. It just isn’t fixed.
Here is the uncomfortable thing about a photo: it is completely accurate and almost useless. It tells you precisely what someone looked like for one five-hundredth of a second, and nothing whatsoever about what they are like to be loved by.
So the first decision — the one that gates every other decision — gets made on the least durable information available. Everything after it is recovery.
We are not going to pretend attraction doesn’t matter. It does, and a product that designs around that is lying to its users. But we don’t think it’s fixed. People become more attractive as you understand them. People become less attractive as you understand them. Almost nobody’s first impression survives contact with who they actually are. If that’s true, then the order you take things in matters enormously — and Seam bets the entire product on it. Understand first, and let attraction do the thing it reliably does, which is follow.
Two people who share the same character architecture will figure out the surface-level stuff. The reverse leaves it to chance.
What we read
None of it is visible in a photo.
When you join, you don’t fill in a form. You have a conversation — about fifteen minutes, scenario-based, closer to a late-night talk than a personality test. It follows what you actually say rather than marching down a script, and it never asks you anything head-on. Direct questions get performed answers.
Out of that we read character. Not traits you’d put on a profile — no hobbies, no job title, no five flattering adjectives. Something closer to the grain of a person: where your sense of worth comes from, what you do with an uncomfortable truth, whether goodwill is your default or something a person has to earn.
We don’t publish the list of what we look for, and that’s deliberate. It’s the one place we’ll ask you to trust us: if you knew exactly what was being measured, you’d stop answering and start performing. Everybody would — it wouldn’t even be dishonest. The questions only work while they’re oblique.
What sits underneath them isn’t software’s opinion of you. What counts as character here, which parts of it actually decide whether two people fit, and what we refuse to look at — those were arguments we had and settled before a line of this existed. Everyone is then read against the same answers.
What we can show you is the shape it makes, and the kind of thing we ask.
- What do you do once you realise you were wrong — and how long does that stay with you?
- Is goodwill your default, or does someone have to earn it?
- Where does your sense of worth come from when nobody is watching?
- Is disruption growth, or is it threat?
- Are you still genuinely curious, or mostly decided?
Your profile isn’t a score and it isn’t a ranking. It’s a shape, and it is only ever compared to other shapes.
What it actually looks like
Four questions, and what they found.
Below is a representative exchange from the opening few minutes. Watch what the questions are doing — not one of them asks this person what they want in someone else, and not one of them could be answered with a preference.
Let’s start somewhere easy. Think about the last time you felt genuinely proud of yourself — not the kind you’d post about, the quiet kind.
What happened?
I turned down a promotion last year. More money, a title I’d wanted for a long time. But the team I’d have been running was being set up to fail and I didn’t want to be the one holding the knife. Nobody really knows I did that.
There’s something telling in “nobody really knows” — you kept the hardest part of it to yourself.
Say that decision had gone the other way, and six months in you realised you’d made the wrong call. What’s the first thing you’d do?
Sit with it for way too long before saying anything. I’d know within a week and still take a month to admit it out loud. Not because I’d be hiding it — more that I’d want to be certain before I made it everyone’s problem.
That gap between knowing and saying sounds less like avoidance and more like not wanting to hand other people your uncertainty.
Different question. Think of a friendship that went quiet — no fight, just drift. Who reaches back first?
Me, usually. Sometimes embarrassingly fast. I find it hard to let something just end by accident.
“End by accident” is a good way to put it. Most people let far more go than they’d admit.
Last one for now — when something does end badly, how long does it live in you?
Longer than it should. I’ll replay a conversation from three years ago at 2am. It’s not a grudge, I just can’t put it down until I understand what happened.
What those four questions produced
Measures themselves by their own ledger rather than an audience — the thing they’re proudest of is the thing nobody was told about. Slow to speak on a hard truth but never evasive about it; the delay is a form of care, not avoidance. Reaches back first when something drifts, and finds it almost impossible to let a relationship end by default. Processes conflict deeply and for a long time, driven more by a need to understand than a need to be right.
How we match
We introduce people who are alike.
Opposites attract is good advice for an evening and bad advice for a life. Seam matches on similarity rather than complement: we look for the people whose grain runs the same way as yours. Tribe-finding, not puzzle-piecing. Two people who both live from the inside out, who both go looking for the hard truth, tend to feel immediately and inexplicably safe with each other — and that feeling turns out to be load-bearing.
There is one exception, and it’s the most interesting thing in the system. On emotional depth — how long something stays with you, how much room it takes up — we don’t score you on your own at all. We score the gap. A deep processor and a light carrier can admire each other for a year and never quite reach each other. Neither one of them is wrong. The distance is the problem. So it’s the one dimension where what matters isn’t your answer, it’s the difference between your answer and theirs.
Exceptional alignment
Eighty-four percent is a number, and a number on its own is worthless. So it never travels alone.
You both keep your own ledger — the things you’re proudest of are the things you never announced. Neither of you can let a relationship end by accident; you both reach back first. And you both carry things for a long time, which means neither of you will be bewildered by the other going quiet for a few days after something hard. That last part is rarer than it sounds.
What we left out
The absences are the product.
No photos
Not hidden until later — absent. The moment a face is on the screen it does all the deciding.
No swiping
Swiping is a machine for making snap judgements quickly. We don’t want fast judgement, so we didn’t build the machine.
No feed
There’s nothing to browse and no infinite list. A small number of real introductions, each with a reason.
No number without a why
Every score comes with the sentence that explains it. A percentage on its own is a slot machine.
Where it stands
You can test the premise with one person.
Seam opens city by city, because a matching pool is worthless until a couple of hundred people are in it. But the premise doesn’t need a city. Send a compatibility check to someone you already know — a friend, someone you’re seeing, someone you’ve known for years. They answer the same questions, and you both get a report on how your characters actually line up.
It works with two people. It is also, frankly, the fastest way to find out whether any of this holds.