In the US this is precisely the traditional approach. The country was built by immigrants who would arrive in an area -- San Francisco or New York for example -- decide that if they didn't want to work in a sweat shop they would have to start a business, then look around to see what their neighborhood needed, and build it. Laundromat, restaurant, feed store, etc.
They didn't need to be intrigued by a problem or have a background that set them up to solve one. They decided to start a business and began by figuring out in what way their region was underserved, then built something to fill the gap. Voila.
There's no obvious reason this approach couldn't work in tech as well.
One, I think starting another instance of a well-known kind of business is very different than creating a new product, especially when one has used that kind of business a lot. So even if what you say were true, I don't think it would be relevant to, "I've started a business, what kind of product should we make."
But two, I'd like to see some evidence that people really did entirely arbitrary things based on demand, without relation to other factors. If we look at a well-known example, the way Indian-Americans have dominated the motel industry, there's pretty clear evidence that wasn't just driven by looking around the neighborhood to see what was needed. E.g.: https://blogs.wsj.com/indiarealtime/2012/06/11/why-indian-am...
Sure, but they were literally neighbors back then. The world has changed since then and we don’t necessarily have neighbors from a completely different walk of life. Additionally, non-tech work has a different domain of deep knowledge necessary to properly serve industry knowledgeably.
Just because your neighbors a welder, doesn’t mean you can whip out a materials inventory app for welders everywhere.
They didn't need to be intrigued by a problem or have a background that set them up to solve one. They decided to start a business and began by figuring out in what way their region was underserved, then built something to fill the gap. Voila.
There's no obvious reason this approach couldn't work in tech as well.