In fact, you probably should barely be thinking about asset development at all, this early on. Unless it's your main area of expertise and you can just crank them out as quickly as any placeholder assets anyway, it's more important to get a working prototype together as quickly and simply as you can, so that you can give it a spin and figure out what needs work.
When I was in the Team Fortress 2 mapping community, it was a near-universal standard to block everything out using plain development textures (that looked like a cross between graph paper and Tyvek sheeting) and iterate on the layout many, many times before even starting work on turning it into something that actually looks like something. And that's just environment design for a game whose mechanics had been locked in years ago. It certainly pays to have a vision and draw up some concept art, but you never know when that vision won't hold up to real-world testing, and it's a lot easier to make sweeping changes to what you've built when it's all just a bunch of nondescript flat surfaces than after you've gotten used to it looking like a haunted house or a pie factory or whatever.