I think one of the points the author is getting at is that there is a lack of (enterprise-level?) tooling around wireguard at the moment. Something that supports 2fa to connect would be nice, etc.
Maybe that is working at the wrong level? Maybe it’s possible to build that feature atop WireGuard, without failing open (potentially catastrophic if you are using it as a VPN).
This is the same in TLS setups (including OpenVPN) and in S/MIME (where it's a reminder that S/MIME probably is only delivering theatre and not any real security to your users).