The OnePartners and investors
Why this company exists
The world knows how to measure almost everything. Almost nobody systematically measures why two people manage to build a life together, and why others do not.
We built a personal matchmaker who holds a real conversation, builds an understanding of a person from 108 research constructs drawn from 17 relationship researchers, and matches with mathematics, not with a swipe. She is live now on Telegram, and you can talk to her from your phone this second.
What we promise: that you will not spend months on someone who could have been ruled out in advance. What we do not promise: love, a wedding, chemistry, "the perfect match".
We are not looking for the perfect match. We reduce uncertainty, and that is a goal you can measure, prove, and get better at.
What the product promises, because that is what gets measured
A relationship is an outcome. It takes months, and it is not entirely in our control. These three promises are a process, they are measured in weeks, and they are in our control.
| Moment | The pain today | The promise | What we measure |
|---|---|---|---|
| Before | The jungle: endless filtering, a sea of profiles | One person, with a reason | Minutes a person invested per week, against ninety minutes a day on an app |
| During | "They wasted my evening" | We will not | "Was this evening worth your time?", asked separately from "want a second?" |
| After | No feedback, and back to zero | We continue from where we stopped | Whether the third suggestion felt more listened-to than the first |
A date that led nowhere and was not a waste is a success here. No app today can produce that sentence.
What accumulates, and why it cannot be bought
Code gets copied. A model gets swapped. An interface gets cloned. An algorithm gets rebuilt. A body of knowledge built from decisions does not.
A swipe is a decision with no reason attached. An app can see that a thumbnail was dismissed and will never know whether it was the height, the town, the phrasing or the shirt. Here every introduction produces a decision, and every decision produces a sentence, because asking for one only feels natural inside a conversation someone is already having.
And the yes matters at least as much as the no. A model built only on rejections converges on the person nobody objects to. The yeses are what turn filtering into direction.
What people say they want does not predict who they choose. What does predict sits at the level of the pair, and is two to three times stronger than anything about an individual (Joel et al. 2020, PNAS, 43 longitudinal studies, 11,000+ couples). And pair-level knowledge cannot be bought in any book. The only way to acquire it is to watch real pairs meet, which takes time money cannot compress.
They measure clicks and screen time. We measure predictions against reality.
We did not write a questionnaire. We encoded the research.
Seventeen relationship researchers, among them Gottman, Helen Fisher, Sternberg, Mikulincer and Shaver, Esther Perel, Aron and Rusbult, taken apart into 108 measurable constructs, each with a definition, a way to measure it and a citation. From them, 191 original interview questions and 79 scoring rules that turn an answer in a conversation into a compatibility vector.
Jane does not invent questions. She draws from the bank by stage of trust. And the matching itself is mathematics, not a language model: zero marginal cost, and auditable.
A panel of practising matchmakers and clinicians rules on what the system is allowed to believe. A rule that cannot be given a test is kept as an opinion, and moves no weight.
The constructs were derived from public sources and rephrased in our own words. The questions are original. No protected measurement instrument was copied.
What is live today, and what is not
Zero revenue. Pre-launch, and not pretending otherwise.
| Live | Not yet |
|---|---|
| The conversation: a short joining form, then a real conversation, one subject at a time | Introductions between two real people. The engine runs and is tested on a synthetic population, and waits for the liquidity floor |
| The Vibez: a personal story page at its own address, with a QR, that can be taken down at any moment | Identity verification against an outside provider. The flow is built; the provider is not chosen |
| Full deletion on one word from the member. An 18+ gate | Terms and privacy documents. With counsel, and we do not launch to strangers without them |
| The pilot is free on purpose: we are buying signal, not revenue | Payment. The infrastructure is written and waiting |
The model
The intended model is a monthly subscription, opt-in after the first introduction, so you pay only after you have received value. The price is deliberately unset: match quality depends on local density, and charging before a city reaches liquidity is charging for something the product cannot yet deliver.
The market
Dating apps are shrinking for the first time. Human matchmaking is growing. People did not stop looking, they stopped looking there.
- On the order of 350 million users worldwide (Statista, 2025 forecast).
- 2025 was the first year the market shrank: down 1.1% to $6.07B.
- Match Group payers down 5% in Q4 2025 (SEC filings).
- Bumble lost 21% of payers in Q1 2026 (quarterly report).
In March 2026 one of the largest dating companies in the world announced its own AI matchmaker, one that holds a private conversation and suggests matches with an explanation. That is the best news we had all year: the market we are betting on stopped being a bet. What it has not yet done, and has no public proof of knowing how to do, is the hard part: understanding two people and predicting the dynamic between them. It measures a person. We measure a pair.
Israel is the first city, not the story. Jane is built so that region, language, currency and legal regime are properties of a member, not constants of the system.
Capital
Capital is not needed to find out whether this works. The product is built and burns very little. Capital buys distribution, the team that turns one working city into many, and the speed to be first.
What we say before you ask
An uncomfortable truth said first becomes a strength. The same truth discovered without us becomes a reason not to invest.
- Three of our 19 sources are tied to a large dating company, and 22% of the constructs lean on them. Everything was built from public material; nothing was given to us privately.
- We are not the only ones without an app, and we do not say we are.
- We have no real dates yet, so we make no accuracy claim about the engine. Calibrated to the literature, not yet to reality.
On the apps people write to be chosen, so they embellish. Here nobody reads the conversation, so there is nobody to write for. That has not been tested in research, and we are the only place that can test it.
Who is building this
A small team in Tel Aviv: a founder, an engineer who rebuilt Jane from scratch, a designer who built the Vibez visual language, and a panel of matchmakers and clinicians who rule on what the system is allowed to believe.
We would like to talk
If you invest, build in this space, or research relationships and want to test this hypothesis on data nobody else collects, write to us.