India’s banks, insurers, and fintech firms are prototyping, testing and deploying a set of AI-infused measures in their customer facing functions. Some of these firms are undergoing a new shift: from software buyers to software builders.
Few people have had a front-row seat to this transformation the way Kiran Jagannath, Head of Financial Services and Conglomerates at AWS India, has. In a wide-ranging conversation, on the sidelines of AWS re:Invent 2025, he explained how cloud, generative AI, agentic orchestration, data modernization, and culture are converging into what he calls “a completely new operating model for financial services.”
From his vantage point, financial service firms in India are at different stages of this shift. Fintech startups are “digital natives, born on the cloud,” and have little legacy to worry about. Insurers, surprisingly, were among the earliest to adopt cloud, beginning around 2017. “Insurance in India is still relatively young,” Mr. Jagannath explained.
“They didn’t have tons of legacy debt, and they naturally gravitated toward cloud because it allowed them to use data to differentiate themselves and innovate quickly.” [Legacy debt in the world of technology refers to old software and hardware systems that are unsuited for current generation of AI workloads.]
A different story
Banks, however, tell a different story. They began their cloud journey later, but once they started, they accelerated faster than any other segment. According to Mr. Jagannath, the turning point came during COVID-19. “When branches couldn’t function as usual, banks had to scale digital channels within weeks. Historically, a project of that size would have taken quarters or even years. Cloud changed the game.”
A second catalyst was the explosion of UPI. Mr. Jagannath noted that its unpredictable, near-doubling annual growth has made local banks rethink procurement models entirely. “You can’t estimate volumes five years out. Sometimes you can’t estimate them two years out. With that kind of volatility, auto-scaling on cloud becomes essential.”
He believes that the cloud question is settled. “The question is: ‘What’s the right architecture and the right migration path?’”
That shift has paved the way for the next frontier: generative AI and agentic AI. Early use cases were task-specific like reducing payment failures, improving fraud detection, accelerating customer service. “The future is agentic AI,” Mr. Jagannath said. “Not one agent, but a multitude of agents with human-like reasoning, working together to complete processes end-to-end.”
He pointed to document-dense workflows like KYC, merchant onboarding, and mortgage origination as early beneficiaries. “Anywhere humans are doing non-value-added tasks — data gathering, validation, submitting to systems — agents are ideal,” he said. “Employees get to focus on value-added activities instead of grunt work.”
A fundamental rethinking
But agentic transformation cannot be layered on top of outdated systems. Legacy cores, monolithic apps, and inconsistent data architectures are real constraints. Acknowledging the contrainsts, Mr. Jagannath said, “For agents to be truly powerful, the underlying systems must be modern.”
“Cloud is not just moving applications from one place to another. The real value is in cloud-native architectures — micro-services, auto-scaling, resilient decoupled systems. Banks are already modernising their cores, whether through hollowing out the core or building dual engines.”
Just as important as technology is culture. Agentic AI systems need a fundamental rethinking of how teams work. The old workflows will be broken down and new ones be built. Small, techno-functional groups will build entire agentic workflows and business and technology groups will merge in this new-age, according to Mr. Jagannath.
A banker in an agentic world
When asked what a banker’s job will look like in an agentic world, Mr. Jagannath was cautious but optimistic. “It’s still evolving,” he said. “Agentic AI is not even a year old as a real industry movement. But I don’t think a digital banker on your phone is far-fetched. We’re already seeing research assistant agents that use customised model. The pace of change is extraordinary.”
He sees the most visible shift in the next six months will be in user experience. Navigation-heavy mobile apps may give way to voice-driven interactions, where customers simply speak commands while agents handle the backend logic. “Banking will become contextual,” he said. “You could be shopping, and within that experience you’ll check your balance or complete a payment without leaving the flow. The banking experience will be embedded in whatever the customer is already doing.”
Published – December 22, 2025 08:10 am IST
