Real-Time Rails, and the AI Payments Shakeup

Opus Advisory Board Member Arvind Ronta had earlier explored where AI is already delivering measurable value across payments, from fraud prevention and customer experience to authorization optimization and agentic commerce. The next question is equally important: How do financial institutions turn these early successes into enterprise-wide business impact?

In this conversation, Arvind shifts the focus from AI use cases to AI execution. Drawing on decades of experience leading payments innovation at Visa and other global institutions, he shares what separates organizations that successfully scale AI from those that remain stuck in pilot mode.

The challenge is becoming more urgent as payments continue to evolve. Real-time payment rails, embedded finance, and agentic commerce are changing how money moves, while AI is reshaping how decisions are made at every stage of the payment journey. Success will depend on more than deploying better models. It will require modern data foundations, redesigned workflows, and deep domain expertise to operationalize AI at scale.

In this conversation, Arvind shares practical guidance on where payments leaders should focus their AI investments, the misconceptions that slow enterprise adoption, and the strategic priorities that will define the next phase of AI-led transformation.

Question: What role do you see AI playing in accelerating innovation around real-time payments, embedded finance, and digital commerce?

Arvind: AI will be foundational to the safe scaling of real-time payments and embedded finance. When you move to instant rails, transaction volume and velocity explode, and traditional, batch-oriented risk and compliance processes cannot keep pace. AI can provide real-time risk monitoring and decisioning at that new tempo, allowing institutions to manage fraud and credit risk dynamically while still offering the immediacy customers expect.

On the commerce side, I think agentic e-commerce is one of the most exciting developments. We are already seeing pilots in which AI agents help consumers shop, compare options, and even initiate transactions, with humans remaining in the loop for higher-risk decisions. McKinsey estimates suggest that between $3 trillion and $5 trillion of online commerce could flow through these agentic channels in B2C retail by 2030.

There is also a convergence among AI, B2B payments, and cross-border flows, particularly where AI meets stablecoins and new settlement mechanisms. That is probably a topic for a deeper dive on its own, but the pattern is consistent: as money movement becomes more programmable, you need smarter, more autonomous systems to orchestrate it. AI is the orchestration layer that will span real-time payments, embedded finance, retail commerce, and, eventually, more complex B2B flows.

Question: Many organizations have AI pilots underway. What separates the companies that are achieving business value from those still struggling to scale?

Arvind: One of the key findings from recent AI research is that while a large majority of organizations report using AI (in some surveys, more than 80%), only a small fraction have scaled it beyond pilots, with only 10% scaling it any function. There is a gap between experimentation and enterprise-wide adoption. In my view, three things separate the leaders from the rest.

First, the leaders are willing to redesign workflows rather than simply “speed them up” with AI. You can certainly sprinkle AI throughout an existing process, but that will not unlock the full potential or scale easily. The real step-change comes when you reimagine how work gets done, whether that is dispute, fraud, or risk management, or onboarding, and then design tools and workflows around an AI-native architecture.

Second, they treat data as infrastructure. In payments, transaction data spans multiple systems, from authorization through clearing, settlement, and servicing. AI needs access across those silos to generate high-quality signals and models. Without that, your data architecture becomes the bottleneck: you may have a great model in one silo, but you cannot scale it across the value chain.

Third, they bring in domain-specific implementation expertise. A generic systems integrator can stand up models and run an impressive proof of concept, but scaling AI in payments means integrating those models into real-time authorization flows, 3-D Secure authentication, dispute rules, and compliance logic. That is where specialist partners who understand both the AI stack and the nuances of the payment domain become critical to moving from pilot to production at scale.

Question: What are the biggest misconceptions leaders have about AI adoption within the payments industry?

Arvind: One big misconception is treating AI as a purely technological project within the IT organization. AI at scale is really about an operating model change. If you simply use AI to wrap legacy platforms and processes, you are unlikely to see transformative value, and you may increase complexity.

A second misconception is “our data is good enough.” In payments, your transaction data is fragmented across multiple systems, and AI models need cross-cutting access to generate strong, reliable signals.

The third misconception is the belief that “we know our data, so we will build everything ourselves.” In my experience, the fastest path from pilot to production is almost never a purely internal build; it is usually a hybrid approach that combines internal strengths with external partners who have done this before and have already navigated the integration, compliance, and testing challenges. That is particularly true in regulated spaces like payments, where the cost of getting things wrong on fraud, privacy, or compliance is very high.

Question: If you were advising a payments executive team today, where would you recommend they focus their AI investments for maximum business impact?

Arvind: If I were advising a payments executive team, I would start with a simple question: Where is AI most likely to move the needle on revenue and risk over the next three to five years? In my view, three areas stand out: fraud and risk decisioning, agentic commerce, and real-time payments.

Fraud and risk are the obvious priority because they sit at the intersection of customer trust, regulatory expectations, and direct P&L impact. Investments here can reduce losses, improve the customer experience (fewer false declines), and enable more confident growth in new products and segments.

Agentic commerce is the second area: as more consumers and businesses use AI agents to discover, compare, and initiate purchases, there is a clear opportunity for payments players that can build the identity, authorization, and settlement rails for these agents.

The third area is AI for real-time payments and embedded finance. As instant rails and embedded experiences proliferate, institutions will need AI to handle real-time risk scoring, sanctions screening, and anomaly detection at scale.

Behind all of this, I would also recommend a horizontal investment in data and workflow modernization. Start treating data as infrastructure and redesign key workflows with AI in mind so that each new AI use case is not an isolated pilot but part of a coherent, scalable architecture.

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

Arvind Ronta is a veteran payments executive and Advisory Board member at Opus Technologies. With over two decades of experience at Visa, Discover/Capital One, and Deloitte, he has led global product strategy in embedded finance, tokenization, and next‑gen credit and installment solutions that power leading fintech and Big Tech payment experiences. He has helped build the FI-led models behind Apple Pay Later, Apple’s BNPL service, and Amazon (moving billions in payment flows). At Opus, Arvind works closely with the leadership team to align AI, real-time payments, and risk decisioning with the company’s core strengths in engineering execution. His role helps banks, networks, and fintechs turn emerging technologies into scalable, compliant products that move the needle on revenue and customer experience.