We use it for an application that aggregates data for consumption by several different teams that all consume different subsets of the data. When you have a pretty simple use-case it's really not that bad to get a decently functioning API off the ground, and because it's self-documenting we can spend our time on more mission critical work.