At the intersection of artificial intelligence and financial technology, Ramandeep “Raman” Aulakh is helping reshape how millions of people interact with money every day. As Director of Product Management at Visa, with previous leadership roles at PayPal and Amazon, Aulakh has spent over a decade building the invisible infrastructure that powers modern commerce. His PayPal bank partnerships platform now runs silently behind every major US bank app, facilitating seamless payment connections for millions of consumers.
In our conversation, Aulakh discusses the quiet AI revolution already transforming payments, why autonomous commerce is closer than we think, and how the industry is preparing for a future where personal agents handle our financial decisions while we sleep.
You’ve built payment platforms used by millions through your work at PayPal and now Visa. For those new to fintech, can you explain how AI is already changing the way we pay for things, often without us realizing it?
The payment industry has quietly relied on machine learning and artificial intelligence for a long time, particularly in fraud prevention—a critical function in financial services. Every time a consumer initiates a payment, AI models are working behind the scenes to block suspicious transactions and approve legitimate ones. With the advancements in AI, fraud prevention models have become more powerful, supporting complex risk decisions in milliseconds while analyzing hundreds of datapoints. The payment ecosystem is able to support smarter transaction routing for faster payments with higher approval rates. Additionally, more and more consumers are getting personalized payment experiences without them realizing that underlying AI models are driving the experience.
Your PayPal bank partnerships platform is now integrated into every major US bank’s app. What does it take to build financial infrastructure that operates at that scale, and how is AI helping these systems become more intelligent?
Building financial infrastructure at scale starts with a few core principles. A successful payment system is designed with security and compliance embedded from day one, with real-time monitoring built into the system’s DNA. To support scale, the systems have to be standardized at the core; however, to support interoperability across a wide variety of systems the edges need to be flexible—picture a “thin, rigid core” and a “thick flexible edge.” Availability is foundational to any payment system given any downtime can cost millions of dollars in revenue. For instances when there are issues, smart error handling and retry strategies are crucial.
AI is bringing flexibility to payments infrastructure, making it more dynamic and adapting. The infrastructure decisioning whether it is access, connectivity, fraud, routing or personalization can all rely on AI models that self-optimize. For instance, AI models are becoming foundational to all risk decisioning, error handling and retry strategies given the wide and deep context these models can support.
At Visa, you work on initiatives with household name tech giants. How do you see the relationship between traditional financial companies and big tech evolving, especially as AI capabilities advance?
I believe the boundaries of traditional roles of financial services providers and tech giants are blurring as co-innovation continues to flourish. Rather than operating within their lanes, companies are building shared technology layers that combine the scale, trust and regulatory experience of financial institutions with speed, personalization, user-centric design of tech giants. With the movement towards autonomous payments through Agentic Commerce, trust and explainability needs to be solidified to win both consumers and regulators. We are seeing industry-wide partnerships across tech companies and payment providers racing to bring compelling AI-powered experiences to consumers.
Buy Now, Pay Later (BNPL) has exploded in popularity. From a product perspective, how is AI being used to make real-time credit decisions, and what misconceptions do people have about how these systems work?
While approaches vary by provider, AI is central to modern BNPL providers both in real-time risk decisioning and in personalizing offers. For risk decisioning, AI models would take into account aspects such as soft credit pulls, identity, device signals, purchase history and any other data sources available to the models. For personalization, purchase history, merchant data and order details can be used to present an optimized BNPL offer to the consumers in milliseconds. Additionally, any activity by the consumer including what and where they purchase and how they make repayments feeds into adaptive learning of the AI models for future experiences. Generally, BNPL providers are transparent about their business models. However, some consumers may still carry misconceptions and think of BNPL as “not-credit” and be unaware that late payments can negatively impact their credit.
You’ve spent over a decade building fintech consumer experiences. What’s the biggest shift you’ve seen in how people expect to interact with financial services, and how is AI enabling those expectations?
Over the past decade, consumer expectation from financial services has gone from “I can take action with my money” to “My money manages itself, in my best interest” with consumers expecting instant, contextual and frictionless experience not just with banking apps but across all payment and financial experiences. With advancements in AI, we are already seeing consumers getting conditioned to expect outcomes and autonomous workflows. The payment industry is getting ready for autonomous payments via Agentic Commerce and contextually embedded finance, where transactions happen naturally in the flow of life without requiring consumers to switch context at all.
Many people worry about AI making financial decisions about their money. As someone who builds these products, how do you balance automation with transparency and user control?
These concerns apply to all areas where AI automation can be applied but are significantly elevated when dealing with people’s money. The way to build experiences that empower consumers while earning their trust comes down to a few principles. Explainability makes consumers feel in control—ensure that decisions made by AI are explained in human-readable language so customers can understand the reasoning. Financial systems that rely on AI automation would see better adoption when customers decide the degree of automation they want for themselves. When AI gets something wrong, there should be ways for customers to reverse or override actions. Similarly, if there is uncertainty then the agents should check with the customer or temporarily hand over control.
From your experience scaling fintech platforms, what are the biggest challenges companies face when trying to integrate AI into existing financial systems? What advice would you give to teams just starting this journey?
I believe the biggest challenge is not the technical integration but everything that’s required to ensure success of the AI integration. Organizational alignment across core teams such as Product, Engineering, Data Science and Compliance, that are required to integrate AI into the products and services, must be bought-in and working in sync. I believe teams that use AI in their day-to-day are better positioned to embed it more effectively into their core products. Given the financial services industry is fragmented and any given transaction touches multiple players, it is important that all ecosystem players are supporting the enhancements required for widespread adoption of AI. For high performing models, data needs to be accurate in the first place, which requires clean, standardized pipelines that securely transfer data with proper access controls. Additionally, the AI models need to prove fairness and explainability to governance committees.
My advice for teams starting their journey of AI integration into financial systems would be to: design for explainability from day one, think of how you embed AI into the core of your product functionality rather than using it at periphery, and build controls for humans-in-the-loop fallbacks and deploy anomaly detection for low-confidence edge cases.
Looking at your career progression from PayPal to Amazon to Visa, how has your approach to product strategy evolved, especially as AI has become more central to financial services?
In the last decade, and especially in the last few years, the focus for product leaders has shifted significantly with AI playing a big role in how product leaders build winning strategies. I used to think about building experiences based on conditions for customer segments, we used to spend significant time on experimentation, we added intelligence to features and execution was granular. With AI supporting products and getting baked into core product functionality, product leaders can focus more on long-term strategic direction and be confident that intelligence can be leveraged system-wide. We can now rethink products end-to-end knowing that execution and deployment will be much faster and we can deploy self-optimizing models to ensure personalized customer experiences.
The payments industry handles trillions of dollars in transactions. How do you think about the intersection of AI innovation and the absolute need for security and reliability in financial systems?
Trust is foundational for the payment industry, and security and reliability are non-negotiable. Even the smallest security breaches or brief downtime could mean loss of millions of dollars in addition to detrimental reputational risk to the impacted players. AI innovations in the core payment rails must inherit and perform flawlessly at the baseline reliability and security on day one of deployment. This means that the AI implementation must operate within security by design architecture and should be pressure tested in shadow mode for a significant time to establish model integrity and performance. I believe AI can be a security amplifier by expanding the dataset being analyzed while operating at or beyond performance benchmarks.
Looking ahead, what emerging trends in fintech and AI do you think will most dramatically change how businesses and consumers handle money over the next five years?
I believe we will see experiences and workflows significantly transform to become far more seamless, autonomous and personalized. We will see AI take control and optimize actions for the consumers’ or businesses’ best interest. AI will also optimize risk management both for businesses and consumers, ensuring the consumers stay within guardrails and businesses are protected from bad actors. I imagine this would also be coupled with hyper-personalized experiences based on real-time context. We are headed towards a world where personal agents would interact with business agents and routine commerce will happen automatically without consumer intervention, and people will only need to step in where they want to or where human judgment is required.
