Call Center: Combat abuse of agent-facilitated appeasements via AI-backed assessment of new appeasement requests

When engaging a customer support agent, a bad actor might seek to exploit customer care procedures for illicit gain by repeatedly requesting various accommodations - whether the reship of an order or line item, claiming that it's lost, or a refund against items claimed as "damaged" or not received, or requesting a discount toward future purchases. Often times, bad actors will dodge detection by minimally modifying names, or billing/shipping addresses, bypassing a direct match on a search, ie. "John Smith vs. Jon Smiith" or "1234 Main Street vs. 1234 Main   Street vs. 1234 Main St."

Given this, agents and customer support managers might spend significant timing combing through a variety of transaction history searches before approving a new appeasement or managed change.  

A better approach would be if upon a new request for appeasement, OMS would automatically execute an AI-backed assessment of the requesting customer's transaction history, including "fuzzy match" on profile data (where permitted), and automatically flag to the Customer Care Rep in real-time if the requesting customer exhibits patterns consistent with repeat offenders of return and/or appeasement practices, along with links / details of substantiating transactions. 

Same approach could be extended to new requests for exchange or refund to detect strange patterns in return abuse.