Bank payment systems face an awkward transition. They have to support immediate payments, ISO 20022 messaging and emerging forms of digital money while continuing to run infrastructure that cannot fail. For Barry Rodrigues, Executive Vice President of Payments at Finastra, that tension is separating banks prepared to modernise their core payment software from those still extending legacy systems. Rodrigues said the distinction is increasingly visible among the approximately 300 banks using Finastra payment software. More progressive institutions are using changes such as ISO 20022 migration as an opportunity to move towards modular, application programming interface (API)-driven and microservices-enabled architectures. Others are placing translation layers over older monolithic applications, postponing the underlying migration. The business question is therefore no longer simply whether banks should modernise payments. It is whether new tooling, particularly artificial intelligence (AI), can compress the time and effort required to move from ageing platforms to modern payment infrastructure. AI is changing the economics of migration One of the clearest examples Rodrigues gave is Finastra's Configurator Assist. The tool maps the use cases and configurations of a bank's existing payment software, including software from another provider, against Finastra's current platform and identifies the gaps that still need to be configured. Rodrigues said that mapping had typically taken about nine months when performed manually for a new customer. Using AI, Finastra can reduce it to about nine weeks. Some differences disappear because they relate to functions created decades ago that the bank no longer uses; others can be configured outside the core code. That matters particularly for large international banks. Rodrigues said institutions operating across dozens of countries often accumulate locally customised systems over three or four decades, making migration substantially harder. He said shortening the mapping and implementation process allows banks to reach time to value faster and see less risk in the migration process. Finastra is also using AI inside its engineering process. Rodrigues described Impact Analyzer, which examines how a code change could affect the wider application, alongside human supervised code generation. At the operational layer, he put the potential efficiency improvement from the newly released Operator Assist at 30%. These capabilities are being developed for the most modern versions of Finastra's software rather than its older releases. Immediate payments raise the cost of standing still The pressure to modernise is also coming from the changing nature of payments themselves. Batch systems were designed around processing windows; domestic immediate-payment systems increasingly require continuous availability, while cross-border payments are beginning to connect directly into local instant payment rails. Rodrigues argued that this changes the architecture banks need. Resilience and scalability become more important as payment processing moves closer to continuous operation, while modularity allows individual software vulnerabilities to be addressed without changing an entire monolithic application. Digital money adds another reason to create a more adaptable architecture. Rodrigues said stablecoins, tokenised deposits and other digital assets could eventually support 24/7 settlement, but he cautioned that true atomic settlement still requires much broader ubiquity, liquidity and reserves. Many tokenised deposit arrangements today ultimately move the underlying money over traditional rails. Rodrigues nevertheless said value is emerging before a fully interoperable global system exists. Some large multinational companies, he said, are already using stablecoins to net-settle more quickly within available networks, reducing the liquidity they have to leave idle. Bank-to-bank or corporate-to-corporate settlement at global scale will take longer. Modernisation needs a business case, not only a technology case For Rodrigues, the investment decision ultimately has to connect the technology programme to the bank's payments business. He said the strongest relationships involve both the business head responsible for profit and loss and the technology head responsible for implementation, because payments modernisation is normally a multi-year programme rather than a one-year technology budget item. He cited banks using the ISO 20022-native capability in Finastra's Global PAYplus platform that, after moving to the modern version, reported improving their corporate request-for-proposal win rates from about one in ten to one in five. The figure is a Finastra client-reported outcome rather than an industry benchmark, but it illustrates the commercial measure Rodrigues wants banks to attach to infrastructure investment. That is a more demanding test of modernisation than completing a standards migration. If newer architecture shortens implementation, improves operations and makes it easier to add new payment capabilities, the return should eventually appear in operating economics or client competitiveness. The next phase is reducing migration friction Rodrigues said large banks cannot replace decades of payment infrastructure quickly. His argument was instead that the cost of delay rises as legacy software becomes harder to maintain and newer capabilities are built for modern architectures. AI therefore has a practical role beyond adding intelligence to payments themselves. In Finastra's model, it is being used to understand old configurations, accelerate engineering, identify defects and assist operators. The proposition is that these tools can reduce the friction around the migration that banks already know they need to undertake. For banks, the relevant measure will be whether faster implementation and lower migration friction are sufficient to move payments modernisation from a recurring technology obligation into a business investment with demonstrable returns.
