Go solo by default, but know exactly what you'd need a cofounder for

More founders are going solo now, and AI is a big reason. A lot of the work that used to need a second person — writing the code, wiring up infra, thinking through edge cases — you can do with an agent at 2am.

So the honest question isn't "do I want a cofounder", it's "where am I weak in a way that kills the company." If it's distribution, a domain you don't understand, or just the ability to keep going when it's grim — that's a real reason. Being lonely or wanting the pitch to sound better is not.

If you do go looking, don't look for someone like you. Look for someone who has actually done the thing you can't. Check your own network first — someone you've shipped with, someone who's seen you under a deadline. And before you commit, build something small together for two weeks. You learn more from one thing you shipped together than from ten coffee chats. The two things I'd vet hardest are speed and taste: someone who wants to ship in two weeks and someone who wants to polish for two months will grind each other down.

Don't recruit a cofounder because you're scared to be alone. Recruit one because there's something you genuinely can't do alone, and you've found the person who's already done it.