It seems like a lot of autotuners have some jaggedness in them, but what's more ideal is something like that glados voice from portal for characters. After preliminary autotuning, how can a voice be made smooth? Can it only be taken care of during recording? Or is it the quality of the software?
Most "auto-tuners" have a setting for how quickly it moves from one note to the next. Just slow it down.
Like tetsujin said, if you're trying to change a lot, or have to fix something really far off the note you want, it's just gonna sound weird.