Hm.
My two cents:
If you take table tennis seriously and if you want to progress, then what a robot should deliver is: balls, that behave like real balls in a real match.
That's the single most important feature. The newgy 2050 does not deliver that (tried the newgy, too, my dealer has this two models). Has only one head. If you turn the balls faster, there will be more spin in the balls. Long balls with few spins aren't possible. Dead balls aren't possible. That's for play around a bit (OK for me, but not what I intend), serious training is not possible.
Yes, the Amicus Pro cannot be programmed to deliver different spins within one programm. May be it will be able later, but: As you cannot see what spin it is (there's no hand of a human being you can observe), it's questionable for a table tennis robot to give different spins within one program anyway - my opinion. I would have to learn by heart the sequence of spins. Not somethin I like.
For an ambitious player, the purpose of a robot is to automatize and enhance specific techniques, that's where the Amicus Pro helps perfectly. The fact, that you can set up to 6 placement points within one program (and a randomising of these placements, so you never know where balls will land - combined with easy or adventurous topspins, backspins, sidespins and spinmixes - and up to 100 balls per minute): that's giving lots of variety. And fun. And heartbeats. I didn't miss the change of spins one minute.
If you want a real match where you face different spins and placements, get a real partner - that's what I have been told and that's what sounds reasonable to me.