It's already been ocurring for a while now without the summary and analysis by AI.
Automatic audio transcription has been kicking around for a long time now, at various levels of accuracy. The only important distinction of the speaking party is which side of the call (employee or customer) it's coming from, and that doesn't require any complicated analysis, just that your recordings capture incoming and outgoing audio on separate channels (which is how phone calls work in the first place).
As far as consent goes, any time you hear "this call may be monitored for training purposes" and stay on the line, as a customer you have consented to the recording. As an employee they usually just include it in your contract or one of the many things they get every employee to sign. Only really matters from a business standpoint if someone is willing to take you to court over it.
Most call center software/systems have these options built in at this point.