I think a trap a lot of developers get into when trying to upskill with a side project is they try to make a start-up or a company or something new or exciting. There's really no need. This is a practice. Sports stars don't try to invent some sort of new move when running drills, so why should developers?
When trying a learning something, just do something new to you, not necessarily new to everyone. The attempt will be more rewarding than the novelty of it. It doesn't have to be fast, new, shiny or epic or even finished. Just start and you'll learn more than you did if you didn't try.
Heads up: As with any terse bit of advice, there will be exceptions of course, and subtleties and nuance that can't possibly be captured in a Byte of Wisdom. This has been what I've experienced, I'd love to hear yours in the comments too!
You can catch up with all the other bytes at http://bytesofwisdom.com