this is usually called finding product market fit or validation stages and each person will approach it slightly differently, but generally what (in my own opinion) works best is:
1) do something that doesn't scale to see if people need it. want to create a mental health app? start by talking to people in mental health-adjacent subreddits etc.
2) build something that solves your problem, then show it to others to see if they faced the same issue, this is usually more 'fun'