AWS Says Banks Face a 24/7 Payments Balancing Act

The speed at which money can be moved has intensified the dilemma banks face in deciding how to process a specific transaction. With payments arriving through various channels like online banking, APIs, or even AI agents, banks must choose from multiple pathways, including batch and instant options. They must also weigh factors like fraud, risk, customer preferences, fees, timing, and recipient accuracy.

Nilesh Dusane, AWS’s global head of institutional payments, emphasized that the core obligations, such as KYC, payment instruction processing, risk management, and fee determination, remain essential regardless of the payment method. The challenge lies in managing these functions within the time constraints of each transaction. While an instant rail might be available but unnecessary for non-time-sensitive payments, urgent transactions leave banks with less time to complete necessary checks and routing decisions.

Cross-border payments introduce additional complexities, as banks must verify beneficiary information and navigate varying processing windows and time zones. Beneficiary validation might not be universally available, with different providers offering varying levels of coverage across regions. The underlying infrastructure must not only support compute and storage but also enable data analysis and decision-making for fraud checks, risk assessment, and KYC procedures during the transaction's active period.

Banks can gradually adopt these capabilities by starting with specific use cases like instant payments or tokenization, employing an orchestration layer to manage eligible transactions. AI further enhances banks' capabilities by extending its role from fraud analysis and payment operations to initiating transactions through agents.

A payment foundation model treats each transaction as a token, analyzing its sequence and relationships to predict subsequent transactions, providing crucial context for fraud detection. The AI system's ability to scale and understand the broader context of transactions is crucial. Banks must stay updated with rule changes from payment schemes, which can vary by country and change frequently.

Traditionally, banks have manually managed these updates, but AI agents can help interpret and implement these changes more efficiently. Localization remains essential even when using common technology across different markets, as each implementation must adapt to the specific payment rails, standards, data requirements, and operating rules of the jurisdiction it serves.

For AWS, the focus is on providing the underlying cloud infrastructure and tools that enable financial institutions to manage these diverse payment options effectively. Ultimately, banks see payment options as an opportunity to serve customers better and leverage new tools like generative AI and payment foundation models to enhance personalized experiences and decision-making.