What is the best way to do customer discovery when all I have is an idea? I don't want to build a prototype until I know the idea is valid and that it's worth doing.
Read up on Lean Startup from CustDev.com and AshMaurya.com.
write down your hypotheses and assumptions around the idea you have
identify a target market and speak to (ideally in-person) 30-50 people in that market and validate (or invalidate) whether they have the problem you think you can solve.
If they do have that problem, then you can find a way to build a prototype or even just put together an HTML prototype or a Powerpoint deck together and go back to those people and interview them about your solution. If that validates, then you build.
Chances are when you do Problem Interviews you’ll find there are holes in your reasoning. You’ll go back and change your hypotheses and assumptions and do it again.
Bottom line – you absolutely do NOT need a prototype to validate an idea.