On the features list: "Polyphonic aftertouch".
It's a little scary; I thought that was a super common feature and didn't think to check that the one I was ordering didn't support it. I'm glad it does, though.
In the end, I'm a little sour on any "more is better" approach to control. Breath control is really cool, but it's hardly ever used. Organs have had massive polyphonic range for quite some time, but relatively few people are going to learn to play them. Roland has been marketing the "D-beam" for at least a decade now and it's about as much of a gimmick as ever. There were pitch ribbons on synthesizers back in the CS-80's era, and they occasionally reappear, but the standard is a pitch bend wheel, and that probably isn't just because MIDI dictated a pitch bend. Most music gear is fairly pedestrian stuff, marketed to existing audiences with a coat of paint and a dose of hype, but that doesn't mean it can be revolutionized - the design constraints are mostly ergonomic these days. We want play to be easy and immediately gratifying and to simultaneously control every aspect of the sound, and somewhere in there, a compromise has to occur.
For example, the idea of performing everything by sample playback as the article implies misses how music is mostly driven by performance and how most instruments are basically optimal to play on their own, even if their skill ceiling is high. A commercial multi-sampled violin will have better tone than an amateur player, but you still have to perform with it like a violinist, using appropriate techniques, if it's meant to channel "violin." Otherwise, it'll be another synth timbre. That latter issue surrounding skill and performance is a much broader one with no "obvious" technological endpoint, even when considering AI performance algorithms. Humans will want to play along too.