A startup's initial problem is usually what the product should be and how to get the first users. It may sound like those are two problems, but they're not. Ideally you solve them simultaneously, and the process of trying to get users shows you what the product should be.
“Launch Fast” is Paul Graham’s mantra. Move from the idea to a minimally functional product as quickly as possible. Only by getting a product into the hands of customers, even if the product is only a prototype, is it possible to know what customers want.1 Launching fast is how to make something people want.
Randall Stross • The Launch Pad
Defining the problem starts with identifying two things: (1) what you want to achieve, and (2) what obstacles stand in the way of getting it.
Shane Parrish • Clear Thinking

However, a PM who just dives into creating a product without understanding the goals might create something that is radically different from what the user needs.
Gayle Laakmann McDowell, Jackie Bavaro • Cracking the PM Interview: How to Land a Product Manager Job in Technology (Cracking the Interview & Career)
Paul Graham suggests that generic launch can be a solid start for the same reason. Get your product out there, see who seems to like it most, and then reach out to those types of users for deeper learning.