I tend to believe that you can get better results by setting levels + delays + EQs of your subs manually with the aid of a measurement mic. Also, the minidsp will allow you to adjust the delay independently between multiple subs letting you time align them.
Even something as simple as level-matching the subs is done much more accurately if you have each sub's actual response curve. An automatic method that just averages SPL can't ignore all the peaks and dips like a human can.
After level matching, time aligning, and cutting significant peaks... you can run audyssey on top of that afterwards. It's fine. You actually do want to do that because you need to align the L/R/C with your subs still.
IMO you would only get comparable results from xt32 alone if your room is perfectly rectangular with no openings and you can do symmetrical placement of subs (midpoints of opposing walls or diagonally opposing corners). And even then, you have to get the app to add a house curve, because with my ears, I think the default target curve is too flat.
All that said... get a mic anyways. There is no faster and more effective way to verify the performance of your setup and quantify improvement.