For many years, payment processors have competed on uptime, scale, and cost.
Artificial intelligence (AI) is changing those dynamics, pushing providers to demonstrate their ability to help financial institutions make better decisions at the speed of trade while maintaining customer trust.
In a conversation with PYMNTS, Matthew PearceVice President of Fraud Risk Management and Dispute Operations at i2cdescribed an industry in which processing transactions became only one part of the value proposition. Organizations increasingly expect partners who can interpret data, reduce fraud, improve approval rates, and support growth without unnecessary friction.
Transformation begins with changing customer requirements. Conversations that previously focused on reliability and operating costs are now focused on speed of implementation and adaptability, Pierce said.
This expectation extends beyond product launches. Financial institutions want to act on transaction data before opportunities disappear, update risk controls and respond to changing market conditions without waiting through long development cycles.
Advertisement: Scroll to continue
Architecture plays a central role in this effort. Fragmented systems create blind spots that limit the effectiveness of AI, while unified platforms allow information to flow more freely across fraud, disputes and risk management functions, Pearce said. The broader payments industry was moving in the same direction.
AI is also changing how organizations measure success.
Pierce was martyred Search from LexisNexis This shows that financial institutions incur more than $5 in associated costs for every dollar they lose directly to fraud, confirming that operational expenses often exceed the losses themselves.
However, reducing fraud cannot come at the expense of customer relationships.
“If a customer experiences a lot of friction, it’s painful for them, and it’s going to put you at the back of the wallet,” Pierce said, adding that the same thing happens if they get scammed.
This observation exemplifies one of the industry’s central dilemmas. Strong authentication protects accounts but may discourage legitimate purchases. Relaxed controls may result in approvals being maintained while increasing financial losses and customer dissatisfaction.
The goal is precision, not severity, introducing friction only when evidence suggests it is justified, while allowing legitimate transactions to proceed uninterrupted. Thus, conversion rates, fraud losses, and customer loyalty become interconnected metrics rather than separate performance indicators.
Models must adapt as quickly as scammers
AI also shortens the useful life of fixed fraud rules. Peirce distinguished between routine decisions that can be automated through solid logic and adaptive systems that continuously learn from new transaction patterns.
i2c brings together data scientists and fraud analysts in continuous feedback cycles designed to improve models while maintaining high approval rates, he said. The goal is to reduce false positives without sacrificing fraud prevention.
This philosophy reflects a broader industry trend. Fraud organizations are increasingly realizing that regular model updates cannot keep up with criminals who are rapidly adopting new techniques and techniques.
Success depends not so much on creating perfect rules as on building systems capable of learning from new information and adapting before emerging schemes spread widely.
Agentic AI introduces new quests
The next challenge may come from purchases initiated by software rather than people.
As agentic AI takes on a larger role in commerce, Pierce said he believes authentication alone will not resolve every dispute.
“What is the intent of the purchase?” he asked. “Did the customer really intend to buy that?”
Determining intent could become one of the defining issues for the next generation of payments. The credentials may be authentic and valid codes while the underlying authority behind the transaction remains uncertain.
Fraudsters are already aggressively adopting AI, Pierce said, and financial institutions must develop equally sophisticated defenses.
Transparency is also important, he said. Organizations need to understand and explain how AI arrives at decisions, especially when automated systems impact approvals, denials, or account actions that directly impact consumers.
These governance questions are likely to become more pressing as AI agents become more actively involved in commerce.
Pearce said he expects processors to contribute information along with infrastructure by the end of the decade. Real-time product configuration, constantly updated fraud controls, and dynamic customization may become standard expectations rather than premium capabilities. Shared behavioral insights and interpretable models can be as valuable as the transaction processing itself.
“Speed, accuracy and transparency have to come together,” Pierce said. “If any of those slip, the process really gets derailed.”
a witness Full interview With Matthew Pearce to learn more about:
- Why are processors increasingly judged on intelligence, adaptability, and speed rather than transaction routing alone?
- How fraud teams use AI to improve approval rates while reducing false positives and maintaining customer loyalty.
- Why agentic AI could force banks and payment providers to rethink consumer intent, governance and accountability before 2030.
