That's the part banks do not care about so I think it's largely irrelevant in the context of this article.
I have no idea where you got that idea, but it's not true. Which is why any deal that happens between institutions (and sometimes intra-institution) has an army of middle men who are responsible for verifying, monitoring, reporting, etc etc.
Imagine you wanted to build a Hubspot/Triggermail/Vero type system, where you send emails to users when certain events are triggered.
Now suppose you use Sendgrid or Amazon to actually transmit your mail. You are unlikely to give very many developers actual sendgrid keys or access to call sendgrid. Most developers will call an internal API to send mail. This internal API will then perform a bunch of sanity checks - duplicate detection, rate limits, dirty word filters, etc.
Can you "cut out the middle man" by letting each internal service have their own internet connected SMTP server, and stop paying sendgrid and fire whoever builds the internal API? That would be pretty silly.
Similarly, a bank can't really get rid of the risk checks, money laundering checks, data normalization checks, etc.