I share some of the concerns, but agree a lot seem to be overplayed.
Biggest one for me is the reliance on a phone app, which may well have a lifespan of maybe 5 years after end of production.
Phones throw out backwards compatibility so readily it would take continuous active effort from KEF to keep it working, whereas a PC control I'd trust to last far longer. Or an integrated web interface would be ideal. Or, at a minimum, a published set of telnet-type control codes, as you see for integration in receivers. Then anyone could knock up a replacement control interface, rather than being reliant on the app.
I see no real reason to be any more worried about their reliability than any other modern amplifier. But when they do fail, if not repairable, you're potentially throwing away a much bigger and more expensive piece of kit. But on the other hand, given the integration, and the potential financial loss, there should be more people willing to pay for repair, hence making the economics of providing the repair more viable. Things are generally hard to repair because they're cheap to replace. Something not cheap to replace should hopefully have more chance of repair.
(It would be nice to see internals - are the amps separated from the control board sufficiently that a third-party amp replacement in the distant future might be conceivable? Or is it so integrated that would never fly? I would expect an amp to fail ahead of any of the digital bits).
Lack of nicer integration into multichannel isn't ideal, but that's really a complaint about lack of icing on top, and it's not really KEF's fault. You can just use the analogue inputs from pre-outs fine, and I'm sure they'll still beat any passive thing with analogue crossover, despite the extra A->D cycle.
And sure, it would be nice to have a "simple" version without all the apps, but economics wouldn't realistically make it cheaper, any more than AV pre-pros end up cheaper than ones with integrated amps.