Maybe, but using state-of-the-art large language models to solve customer support queries with agentic can quickly use a lot of tokens. What I understood from the talk is that they used agents with limited responsibility and (assumption from me) smaller models, to the make sure the answers were quick, reliable and not too costly.

There are several payments processing companies that are already largely using AI for customer support queries. They still have an escape hatch to a human but at least one of those companies (on the smaller side) is reporting a ~99% success rate, they are down to a handful of human customer service employees now for cases where the customer can't find/produce the transaction ID.