If I understood correctly, you're saying starting small is best because things that may seem small may actually be a thread to something big .. even if it isn't obvious .. but the habit of starting things will be the truest test .. this is a good one.
The best thing imo is build something that addresses a pain you already live and understand. Trying to build for an industry you do not know will be much harder
Instead of thinking about what to build, think about what to solve, so that you are not running around with a hammer looking for nails, and instead you are looking through the toolbox for the best tool for the problem at hand