127 FINANCIAL SERVICES agents capable of executing tasks, orchestrating processes and even collaborating with other agents across the enterprise. In payments, for example, we’re seeing AI agents handle everything from transaction reconciliation and exception repair to dispute resolution and know your customer. These agents don’t just automate – they reason, learn and adopt. They can interpret SWIFT messages, diagnose errors, draft responses and even take corrective action through legacy systems using computer-using agents. “This evolution is truly shaping the frontier firm, a model where humans set direction and AI agents operate the business across three phases: AI copilots assist employees, agents join teams as digital colleagues, and agents run entire workflows with humans in the loop for oversight.” Both Pichach and Ophir warn that the success of financial organisations will increasingly hinge on their ability to harness AI’s potential. “Banks must embrace AI and modernise their core platforms or risk becoming as obsolete as a floppy disk,” says Ophir, who advises firms follow six strategies. “First, they need to rearchitect core systems then they need to invest in data foundations and build unified data estates that break down silos and enable real-time analytics.” Then, firms can implement frameworks for AI transparency, fairness and accountability. “Develop the guardrails, set the right culture and experiment with Copilot and AI technology as it will become the norm for enterprise applications,” she explains. The final three strategies include upskilling the workforce by training them in AI literacy, prompt engineering and agentic workflows; prototyping and scaling use cases by starting with high-impact, low-risk cases like fraud detection and policy intake; and engaging with ecosystems to co-innovate and derisk adoption. “The re-engineering process is a great way to get into the heart of the core business; it’s a science in itself,” says Ophir. “After you’ve built a proof of concept, you should capture the potential time saved and the return on investment. Also, make sure to embed compliance checks where it makes sense to save time and effort later.” Leaders must also prioritise data readiness, composable architectures, and governance and trust. “AI is only as good as the data it learns from,” says Pichach. “Harmonising structured and unstructured data – ISO 20022 in payments for example – is foundational. Legacy systems need to be opened up with APIs and cloudnative services. And, as AI agents take on more responsibility, clear policies around authentication, authorisation and auditability become critical. Microsoft’s work with Mastercard, PayPal and Visa on using tokens to secure agentic payments are examples of how we’re addressing this.” Mastercard’s Agentic Payments Program uses agentic tokens and payment passkeys to enhance trust, security and control across transactions. It is working with Microsoft to integrate Microsoft Azure OpenAI Service and Microsoft Copilot Studio with its payment solutions to develop and scale agentic commerce. The tokenisation technology will enable payments to be initiated through conversational interfaces and conducted by the millions of merchants of all sizes that support online commerce today. “Mastercard is transforming the way the world pays for the better by anticipating consumer needs on the horizon,” says Jorn Lambert, chief product officer at Mastercard. “The launch of Mastercard Agent Pay marks our initial steps in redefining commerce in the AI era, including new merchant interfaces to distinguish trusted agents from bad actors using agentic technology.” With the rapid evolution of AI from a supporting tool to the very foundation of business operations, the imperative for leadership is clear, says Pichach. “The bottom line? AI is no longer a layer on top of core systems – it’s becoming the operating model. Leaders who embrace this shift now will define the next generation of financial services.” “ Banks must embrace AI and modernise their core platforms or risk becoming as obsolete as a floppy disk” DALIA OPHIR, MICROSOFT
RkJQdWJsaXNoZXIy NzQ1NTk=