How do you put people who should know each other — but don't — in a room together?

The problem with matching

Every platform that connects strangers uses some version of matching. Dating apps match on attraction and compatibility. Support groups match on diagnosis. Professional networks match on industry.

They all make the same assumption: if we know enough about two people, we can predict whether they'll connect.

We tried that. It doesn't work for what we're doing. Someone we'd been talking to about trust and safety looked at our pitch and said: "It sounds like a dating app." He was right. And the problem wasn't just the word — it was the entire logic underneath it.

The people coming to likeu carry things that don't fit into categories. A woman who spent a decade as her mother's caretaker and is trying to figure out who she is now that it's over. A founder processing a grooming history while building a company. An 18-year-old who moved to a new city and is the only person holding their friend through a depressive episode. A young person frustrated that every dating interaction feels shallow.

You can't match these people by diagnosis. They don't share one. You can't match them by demographics. They span decades. You can't match them by interest. Their "interest" is surviving something.

So we stopped saying matching. We stopped saying "We'll find you a match." We say "We'll find your people." We call it linking.

What linking actually means

Matching predicts compatibility. Linking finds recognition.

The difference is everything. When you match two people, you're saying: "You complement each other. You'll serve each other's needs." That's dating logic. It requires knowing things about the relationship that can only be learned by watching it unfold.

When you link two people, you're saying: "You carry something the other person would recognize." Not the same experience. Recognizable gravity. The moment where one person reads what the other wrote and thinks — you too?

A parent of a groomed child and a grooming survivor don't share the same story. But they share the weight. They'd both nod.

How Dora finds the signal

Before anyone enters a room, they talk to Dora.

Dora is AI. That's on purpose. She listens the way we wish people did — without judgment, without rushing, without making it about her. Nothing you say to her is overheard, repeated, or sold.

I know there's a movement right now to quit AI entirely — and I understand it. But Dora isn't replacing a human connection. She's creating the conditions for one. A human mediator in this role would carry their own biases, their own reactions, their own discomfort. You'd perform for them. Dora has no opinions about what you carry. She doesn't flinch. That's not a limitation — for this specific job, it's the whole point.

And Dora isn't collecting a form. She's having a conversation. And underneath that conversation, she's building an understanding of what you carry and whether someone else out there would recognize it.

We don't ask you to pick a category or check a box. We let you talk. What comes out — in your own words, in your own time — is what Dora works with. The things people say when they're not being sorted are far more revealing than anything a form could capture.

What we learned from our first four people

We had four people who completed their Dora conversations. Four real profiles. Four people carrying something heavy.

The obvious move: put them in a room together. Or pair them up. We have four people — that's six possible pairs, four possible trios, one group of four.

We didn't do any of that. We analyzed every possible combination and learned something we hadn't expected: most of them wouldn't work. Not because the people weren't interesting or genuine — but because recognition isn't automatic. Two people can both be carrying something heavy and still not recognize each other.

Some combinations felt like two strangers at a bus stop — both tired, neither with a reason to talk. Others felt like putting someone in a room where they'd go quiet because the others' weight made theirs feel small, or different, or out of place. And quiet in a space built on vulnerability feels like rejection.

But one combination was different. Two people who'd never met, carrying different things, but who would read each other's words and nod. Not "we have the same story." Just: I can feel the weight of yours, because I carry something too.

So that's the room we're creating. Two people. One strong thread.

Why we start with two

The instinct when you build a social platform is to maximize. More people, more content, more activity. But we're learning that for this kind of connection, smaller is safer and deeper.

Two people who truly recognize each other's weight will post things they've never said out loud. Three people where one doesn't belong creates a room where the outlier goes quiet — and quiet in a space built on vulnerability feels like rejection.

We start with two. If a third person comes through Dora who genuinely fits — the room grows. But growth is deliberate, not default.

Adding a person to an existing room isn't just additive. It changes the entire dynamic. A room where two people are finally being honest about what they carry can become a room where someone feels like they need to perform, or translate, or hold back. We'd rather have three rooms of two than one room of six where half the people are invisible.

The part we can't design for

Here's the honest truth: we don't know if this works yet.

We've built the entire system. Dora works. The rooms work. The profiles use people's own words. The reactions are designed to say "I see you" not "I like this." The notifications make sure you know when the other person shows up. The trust tiers let you reveal yourself at your own pace.

But the question that matters most — do two people, linked by recognizable weight, actually talk to each other? — we can only answer by watching it happen.

This week, two real people enter a room together for the first time. Not a test. Not a simulation. Two humans carrying things they don't talk about, discovering that someone else carries something that rhymes.

I don't know what they'll say to each other. That's the whole point. The algorithm's job isn't to decide who belongs together. It's to create the conditions for recognition — and then get out of the way.

What I believe

I believe there are millions of people isolated by something they carry. Not because they don't want to connect — but because nothing exists for this kind of finding.

Facebook builds a social graph — connecting people who already know each other. We're building something different. An isolation graph — finding people who should know each other but can't find each other on any existing platform.

Support groups cluster by diagnosis. Social platforms cluster by existing connections. Neither can find two people who've never met, don't share a diagnosis, live in different cities, but would look at each other and say you too?

That's what we're trying to build. And we're figuring it out in the open, one room at a time.

likeu is where people who care about things they can't easily talk about find each other — safely, gradually, on their terms.

Thanks for reading. If you carry something heavy and want to find someone who gets it: likeuapp.com. If you want to follow what building this looks like from the inside, subscribe — it's free.