I think the concept is a really cool idea with some serious potential but there's some things here though that make me skeptical of the implementation.
For a game with this level of graphical fidelity the hardware requirements seem too high. A game this modern should not have as many loading screens as it does. It will make navigation and exploration feel really tedious and I can see players getting annoyed by it after an extended playthrough. If these issues are engine specific then it might be worth switching over to a different one sooner rather later since I imagine it will only get worse the more complicated it gets.
I'm also not convinced navigation needs to be fully 3d with a free camera. The gameplay (so far) boils down to just walking to objects or people and pressing space to interact. This could very easily be done VN-style with mouse clicks instead. The demographic you're targeting would be more used to that format as well and it could make the game more performant in your engine. Alternatively you could try a fixed top-down perspective for walking around.
Rooms are too spacious and should probably be shrunk down. A lot of the walking feels unnecessary.
Games like these are carried hard by their character writing and designs so I really hope you can nail that aspect. So far, none of the designs really stand out to me but I'll chalk that up to them being placeholder vroid models.
I like a lot of the UI but it's very obvious to see their inspirations (i.e. main textbox looks ripped from Danganronpa 1. The choice menu is very obviously persona 5-esque). Not necessarily a bad thing but might be worth trying to differentiate them more for a more unique visual identity. I also don't like that Bianca's textbox look so different from the other ones but that's probably just a personal nitpick.
Hopefully I'm not coming off too harsh. I'm really interested to see how this one develops.