Banks have spent the past several years asking whether artificial intelligence can make better decisions. The more important question is becoming whether their infrastructure can execute those decisions.
While 52% of banks have piloted agentic AI, just 16% have fully deployed production use cases, according to data cited in the September edition of the AI-Ready Issuing Tracker® Series, a PYMNTS Intelligence collaboration with Thales.
As agentic AI moves from analyzing transactions and recommending actions toward participating directly in operational workflows, it is exposing a structural mismatch inside financial institutions. Much of the technology supporting card issuance was designed around people coordinating work across separate systems.
AI agents, by contrast, are designed to execute that work across systems in real time. The emerging competitive gap, in other words, may be less about AI maturity than operational maturity.
AI Is Exposing the Execution Problem Across Financial Services
APIs, interoperable data and orchestration have traditionally been sold as modernization tools that help banks launch products faster and connect systems more efficiently. Agentic AI turns them into the mechanisms through which autonomous decisions become operational actions.
Card replacement offers a useful example. One lost card can require address verification, physical and virtual card issuance, PIN management, token updates, fulfillment and customer communication. In fragmented environments, each becomes another handoff.
A modern issuing platform can instead treat the request as a single lifecycle event, orchestrating those actions across systems in real time. The Tracker found that roughly 25% of inbound contact center calls in some issuing environments involve card status questions. Connecting issuance, fulfillment and communications allows those inquiries to become proactive notifications while creating the same infrastructure an AI agent needs to manage the process autonomously.
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
Banks are no longer simply replacing old technology with newer technology. They are building the execution environment through which AI can produce returns.
Read the Tracker: Why Modern Issuing Platforms Will Determine Which AI Strategies Succeed
Banks do not have to choose between modernizing today’s workflows and preparing for tomorrow’s automation. Done correctly, the same architecture does both. The moat, in other words, is moving down the stack.
AI models are becoming more capable and more accessible. That makes raw intelligence harder to defend as a durable competitive advantage. What is harder to replicate is everything surrounding the model, including proprietary data, institutional rules, permissions, APIs, controls, and the ability to move from decision to action without breaking compliance or customer experience.
A bank with sophisticated AI sitting on fragmented infrastructure may still struggle to produce meaningful operating leverage, while a bank with modern, well-governed execution infrastructure can potentially make every subsequent generation of AI more valuable.
The next phase of AI in financial services will therefore be less about what the technology can think. It will be about how much of the business it can safely execute.
For issuers who modernize now, that is a much more valuable race to win.
For all PYMNTS AI and B2B coverage, subscribe to the daily AI and B2B Newsletters.
At PYMNTS Intelligence, we work with businesses to uncover insights that fuel intelligent, data-driven discussions on changing customer expectations, a more connected economy and the strategic shifts necessary to achieve outcomes. With rigorous research methodologies and unwavering commitment to objective quality, we offer trusted data to grow your business. As our partner, you’ll have access to our diverse team of PhDs, researchers, data analysts, number crunchers, subject matter veterans and editorial experts.
Add as Preferred Source
