How to find a co-founder

Be specific before you look

Most co-founder searches fail before they start, because the ask is unfalsifiable. "Looking for a technical co-founder for an exciting startup" gives a reader nothing to say yes or no to. A specific ask names the market, the stage the work is actually at, the hours it needs each week, the equity on offer and the vesting attached to it. It is uncomfortable to write because it forces decisions you have been postponing, which is exactly why it filters well.

The test of specificity is whether a qualified person could decline for a good reason. If nobody can sensibly say no to your ask, nobody can sensibly say yes to it either.

Look where the intent already is

Your existing network is the highest-signal place and the smallest. People who have seen you work already know your failure modes, which is worth more than any interview. Its limit is arithmetic: most people do not know enough candidates with the right complementary skill and the right appetite for risk at the same time.

Beyond that, the options are accelerators and founder programmes, communities built around a craft rather than around startups generally, and matching platforms. Each has a characteristic failure. Accelerators select for people already committed to their own idea. General startup communities are full of people who want to be in a startup rather than to do a specific job. Matching platforms are efficient at introductions and weak at everything after the introduction — which is why the screening below matters more than where you looked.

Screen for time and conflict, not just skill

Skill is the easiest thing to assess and the least likely to break the partnership. What breaks partnerships is a mismatch in available hours, in risk tolerance, and in how each person behaves during disagreement. Ask directly how many hours a week are genuinely free, and treat an unrealistic answer as a serious finding rather than enthusiasm.

The seats in this ecosystem are a useful illustration of the skill point. Across 99 open CEO seats, 98 name sales among their required skills and 25 name engineering. These are operator roles: the scarce thing is someone who will do commercial work, not someone who can write the software.

Test on something real, then write it down

Do a small piece of actual work together before either of you commits. Two or three weeks on something with a deliverable and a deadline tells you more than three months of coffees: you learn how they handle being wrong, whether they finish, and whether the disagreements are productive.

Then agree the terms in writing before the real commitment starts — equity split, vesting, cliff, roles, and what happens if one of you leaves. Doing this early is not a sign of distrust. It is the one conversation that gets dramatically harder every month you postpone it.

A shortcut that exists here

One reason this search is hard is that you are usually solving two problems at once: finding a person and finding a venture worth their time. Joining an existing venture collapses that into one. In this ecosystem there are 99 ventures with an open CEO seat across 14 industries and 4 countries, 51 of them already in beta and 42 with an MVP ready — so the market, the product and the terms are defined before the conversation starts, and what remains to establish is fit.