Building tools for other companies’ platforms (first one was Twitter, second one is macOS, last is OpenAI) is always a double edged sword.
It makes growth / user acquisition easier, because you’re going where the people already are. But it also means you’re at the mercy of the underlying platform waking up and invalidating your business model (as happened with Twitter).
As you suggest, the only way to survive in this mode is to be continuously building new products around a diverse group of platforms. It helps if they’re not too diverse, because you can cross-sell (i.e. the correlation of Twitter users, macOS users and chatgpt web users is pretty high). When one platform shifts, you find another.
It’s a risky method but not a bad one if your morale can stomach the down turns in any individual platform.
It seems like he needs to constantly push new products out there to maintain a steady income.