Key: Start with a space you're excited about—that's big and underserved—and ideate. But, as soon as you have a concept you're excited about, quickly talk to potential users. @zoink narrowed his focus to two areas he saw potential in (drones and WebGL) and then started digging: