Running a company is the loneliest well-paid job going, and the data on it is worse than the vibes suggest. Michael Freeman's UCSF research found 72% of entrepreneurs reported a lifetime history of at least one mental health concern against 48% of a comparison group, with 30% reporting depression versus roughly 7% in the general population. Sifted's 2025 survey of 156 founders found 46% rated their own mental health as "bad" or "very bad". This is not a soft problem at the edges of founder life. It is the median experience of the job.
The reason it persists is structural. A founder cannot be fully honest with their team, because confidence is part of the product. They cannot be fully honest with investors, because the next round depends on the story holding. They often cannot be honest with a partner who has heard about the runway problem for eleven consecutive months. The only people who can absorb it are other founders, and the mechanisms for meeting those are all badly designed.
The existing options split into two piles, both flawed. Free networking is the problem wearing a lanyard: a room where everyone is waiting for their turn to pitch, optimised for collecting contacts rather than saying anything true. Paid peer groups are genuinely good and genuinely expensive: YPO can run past $20,000 in year one, Vistage around $19,000, Hampton $8,500 to $15,000, and EO, the most accessible of them, $4,400 to $7,000. Those are excellent products for founders doing $3M+ in revenue. They are irrelevant to the person two years in at $40k MRR who is quietly falling apart, which is most of the people reading this.
And the middle is empty. Not because nobody wants it, but because the obvious version of it, monthly coffee, does not create the conditions where anyone says anything real. Here's the idea.
A seasonal club that puts six founders who have never met into one activity a week in the same city, for twelve weeks, with a no-pitching rule.
The sequencing matters. This is a local operations business first and a product second, and every attempt to invert that has failed.
What the comparable actually did