It sounds like your team is fairly young (student-wise, at least), so letting them fail wouldn't be the worst thing in the world, since they'll always have future years to improve. In my experience, the most effective mentoring has been the question-asking type - why are you doing this? Why did you decide not to do this? Teach them to think for themselves by asking them the questions they should have already asked themselves. If they've fallen in love with their designs and won't listen to anything negative about them (trust me, it definitely happens), then at least encourage them to test their design early so that they can fix it.
As someone on the other side of the mentor/student partnership, I can safely say that mentoring is a very tough job. Walking the line of helping too much and helping too little must be a huge challenge, and your students (even if they don't right now) will come to appreciate your efforts.
Good luck!
