I was the technical cofounder of my first business, Wishery. The company did OK: thousands of customers and thousands of dollars/month of recurring revenue toward the end. When we failed to raise capital (more on this tomorrow), my partner brokered a sale to Zenbox and the business was sold.

My partner and I fought a lot. I was young and arrogant, but honest; he was more humble, but also delusional. The company never had a viable business model. But there's another facet to it: our relationship as CEO and "technical cofounder".

What I've realized is, "technical cofounder" is a bullshit job title. This title indicates a CEO who doesn't want to share decision-making, but needs a strong technologist to check some kind of box, usually fundraising—though if they had their way, they'd prefer to hire an employee and be the boss. A relationship where one person wants a partnership (even if there's a leader) but the other wants to be the boss isn't going to last; that's what "technical cofounder" signals, and it's wise to avoid, whether as an investor, employeee, or otherwise. (Sidenote: Y Combinator is smart to require a video of the founders; firsthand experience has taught me most early startup failures happen due to bad team chemistry, and you can learn a lot about team dynamics from watching how people talk about each other.)

One better way: a single CEO with employees, the "Jeff Bezos model". This configuration is stable, but requires someone with a lot of business experience, usually an older founder.

The other model I've seen work is primus inter pares: first among equals. Many professional services firms (law, accounting, architecture) operate this way: a group of partners with one manager/leader whose decisions are final. This model acknowledges that a group needs an identified final decision-maker, but major decisions cannot be taken without support of the broader group. Primus inter pares is very hard to get right: it takes years, and requires people who know each other very well, with a high level of mutual respect. But I think it's the best model for professional services and technology firms, which are more similar than either realize.