Looking at the list, I feel timing makes a big difference*. You need to be early enough that people think you are a bit crazy, but not too early that the tech isn't there or even early adopters are not ready.

Openclaw for example could have been built in 2023, but it did well in 2026. I don't think 2023 was ready for it :-)

* Modulo survivor bias, execution, funding, brilliant fouders, great advisors, pure luck etc.*