I found an area far outside of tech that had very real problems that nobody was really solving; in fact, the existing tools that should have solved this problem turned out to be a nightmare for the intended users. I've been working on it for a few months now and gotten very positive feedback, and have a steadily-growing userbase now. It's more of a side-project right now, maybe 10 hours/week, but maybe someday it'll evolve into an actual business (:
Not willing to share specifics about the specific area I'm doing this in, but the general idea stands. It helps a lot that the problem I'm working to solve is one I personally dealt with, so I understood a lot about what the problem was and was able to effectively solve it for myself. This translated into producing something that a lot of people like and use regularly.
Even more broadly: solve your own problems, then generalise out. If you're solving a real problem that you have, that isn't incredibly specific to something about you/your life/the way you do things/..., then odds are there's probably others out there with the same problem. Every even modestly successful project I've had started with solving my own problems, then polishing the solutions for others to use.