From my perspective. Firstly, know yourself. Try to identify directions of knowledge you have a natural attraction towards. This phase takes time. Use the sense of curiosity as opposed to common entrepreneurial logic. What would you do even if unpaid? Do a few small projects, explore different directions before committing to something specific.

By the time you’ve identified and experimented with what makes you tick you will have already built an idea of what people discuss on forums, etc. There will be already existing solutions and incumbent providers. Don't be intimidated by them. Read what users say and how they express themselves within that domain. Very likely incumbent providers are “old style” burdened by legacy patterns. People use them because there’s no alternatives.

If at this stage you do have a genuine feeling that you can do better - supported by product vision (that's important!) - go for it. But it has to be real for you. You’ll need enough passion to last through the first year of development. If you will survive that year the chances are you will have created something genuine and there’s some real IP behind your effort. Users will recognise that.