Start working on something, literally anything. In the course of working on it, you will encounter problems. Congratulations, you now have problems that you need solving which hopefully other people experience too and would pay for a solution. Now you build that solution and you sell it, abandoning your initial project.
This is what I do at least, it works well. It also encourages more B2B businesses than B2C. Recently I've looked into form handling for businesses, an email API that's good, an uptime monitor for my products, etc.