Two things are true at once here and most articles pick one. Women are heavily underrepresented in the places founders go to find each other, and the funding headlines have never looked better. Neither fact tells you much about how to actually run the search.
What follows is the arithmetic of the pool, what the headline numbers hide, and the method that works on a small pool.
This is not a pub. Ages 13+.
On Y Combinator's cofounder matching platform, 13 percent of profiles are women. Roughly one in eight, before you have filtered for anything at all.
Now apply the filters everybody applies. Technical or commercial, your city, your stage, your domain, available in the next three months. Each one is a fraction of a fraction, and the result on a pool that small is a shortlist of nobody.
Which points at the method. On a small pool, tightening the filter is the wrong move and widening the search is the right one. The people worth talking to are disproportionately not on a founder platform: they are employed, in a research group, running something else, or in a country your network does not reach.
Worth knowing, because someone will quote them at you.
PitchBook put startups with at least one female founder at 19.9 percent of US venture dollars in 2024, $38.8 billion. In 2025 that hit a record 27.7 percent, $73.6 billion. Read alone it looks like a step change.
Underneath, over 40 percent of that 2025 figure came from a handful of enormous rounds at two very large private companies, and teams that are entirely female stayed at roughly one to two percent of dollars, with the capital they raised falling about 22 percent year on year.
So the headline improved and the typical experience did not. The practical conclusion is a narrow one: none of these aggregates describe your seed round. Choose a cofounder on capability and fit, not on a funding narrative, because the narrative is being set by two companies you are not.
Vision conversations are pleasant and predict almost nothing. Noam Wasserman's work on more than ten thousand founders puts 65 percent of high-potential startup failures on conflict inside the founding team rather than on the product or the market, and the recurring causes are roles, relationships and rewards.
So lead with the numbers. How many months of personal runway do you each have. How many hours a week, honestly, next month, including the consulting neither of you mentioned. Who makes the final call when you cannot agree. What a good outcome looks like in five years, given that a founder who wants to sell for twenty million and one who wants to run it forever will make opposite decisions about every round.
Then equity, written down, with a four-year vest and a one-year cliff. The split matters less than the schedule, which is the thing that protects both of you if one of you leaves in month five.
And screen for complement rather than agreement. Somebody who thinks exactly as you do is comfortable and adds nothing you did not already have.
Do not sign anything on day one. Scope one small thing with a real deadline: a landing page with traffic pointed at it, five customer interviews with written notes, a spec somebody outside can read.
You learn what you cannot learn in conversation. Whether they finish. Whether they tell you early when they are going to be late. How they behave at hour nine when it is not working, and whether the disagreement in the middle of it was useful or exhausting.
Coffees select for people who are good at coffees.
On Notapub you write what you are building and the gap in plain words. "Second-time founder, health data, six paying pilots, I do the commercial side and I need somebody who can own the engineering." That sentence is matched to people whose own words line up with it, and they hear about you.
The part that matters on a thin pool is that it keeps working. Your sentence is still being matched next month, so the person who decides in October that she is finally leaving is brought to you then, without you having written anything again. And it is not a founder directory, so the people it reaches were never required to have declared themselves as looking.
If you want to get the ask straight first, you can call. Nora is a voice rather than a person and she says so. Fifteen minutes usually gets the gap out of you in words worth sending. Finishing a profile also pays $50, reviewed by hand.
Alumni groups and friend circles repeat the demographics they started with, and the dedicated platforms run about 13 percent women. On a pool that thin, filtering harder returns nobody, so the search has to get wider instead.
One concrete deliverable with a real deadline: five customer interviews, a working landing page, a spec. Watch communication, pace under pressure and whether they finish.
You say what you are building and what you are missing, and the people whose own words are the other half of yours are brought to you. It keeps matching after you close it. Free on iPhone, Android and the web.