BOSTON — Banks and other card issuers could use artificial intelligence to speed up replacement cards, improve customer updates and reduce avoidable service calls, but those gains depend on modernizing the systems that handle the work, according to a study by PYMNTS Intelligence and Thales.
The findings are part of the report “Why Modern Issuing Platforms Will Determine Which AI Strategies Succeed,” in the AI-Ready Issuing Tracker series. The study says AI is most useful when it can draw on current information, coordinate work across departments and operate within established institutional rules.
Card Replacement Offers Starting Point
The report identifies replacement of lost or damaged cards as a practical starting point because a single request can require several separate actions, including:
- Confirming a customer’s address and validating the replacement request.
- Ordering and producing a physical replacement card.
- Updating digital cards and substitute payment credentials used to protect account details.
- Coordinating the steps across teams while applying the issuer’s security and operating policies.
According to PYMNTS and Thales, a connected issuing platform could allow some of those steps to proceed simultaneously rather than waiting for sequential handoffs, potentially shortening the replacement process.
Card-Status Calls Present Opportunity
About 25% of inbound contact-center calls concern card status, based on observations from Thales’ discussions with customers cited in the report. The figure does not indicate how many calls automation could eliminate.
The study says connecting card production, delivery tracking and customer communications could allow issuers to provide more timely status updates and proactively notify cardholders. That could reduce the need for customers to call while allowing service representatives to concentrate on more complicated issues.
AI Adoption Remains Uneven
The report also cites research showing a gap between testing AI and deploying it widely:
- 52% of banks have piloted agentic AI, according to EY research cited by PYMNTS and Thales.
- 16% of banks have fully deployed agentic AI use cases in production, the EY research found.
- 60% of neobanks and financial technology companies reported high or transformational improvements from AI in transaction risk management, compared with 44% of incumbent banks, according to Visa research cited in the report.
Controls Needed for Expansion
PYMNTS and Thales said reliable customer and transaction data, access to institutional policies and clear human oversight are necessary for AI to perform tasks consistently and securely.
The same infrastructure could support more complex activities, including transaction disputes and launching new card products, which often involve multiple teams and systems.
The report recommends that issuers identify disconnected workflows and modernize them alongside governance and oversight measures, rather than treating AI deployment as a stand-alone technology project.
