Beyond digital banking: Why intelligence must be built into the product

By Mukesh Bangia, Vice President – Product Management Group, Nucleus Software

India’s digital banking journey has been remarkable for the pace and scale at which it has happened. In FY 2025-26, UPI processed over 24,161 crores1 transactions worth approximately ₹314.23 lakh crore. By volume, UPI also represented nearly 49% of global real-time payment transactions in 2025. What was once considered an ambitious vision of financial inclusion has become part of everyday life for millions.

Yet faster transactions do not automatically make banking more intelligent. The next phase depends on technology helping institutions interpret information, make timely decisions and respond as customer needs and risks change.

Banking decisions rarely depend on a single interaction. A loan application can move digitally through a workflow, yet its assessment may still depend on disconnected data and manual interpretation. Digitisation has improved access and efficiency. The next challenge is to make banking operations and decisions more connected, contextual and responsive.

The next step is intelligence across the product lifecycle
For much of the past decade, banking innovation has focused on digitising journeys, automating routine tasks and reducing manual intervention. These changes improve efficiency and access, but automation alone does not create intelligence. In lending, insight needs to travel across origination, underwriting, servicing and collections so changes in repayment behaviour reach decisionmakers in time.

Few financial institutions operate on a completely modern or unified technology landscape. Their environments combine systems built at different times, on different architectures and for different requirements. Replacing the entire estate before adopting new capabilities is neither practical nor necessary.

AI must therefore shape how products process information, apply business rules and support decisions and not sit as an isolated feature. A prediction creates value only when systems translate it into a timely and accountable action. Product teams must connect data, decision logic and workflows in an explainable way.

Global banking is moving in the same direction
Intelligent banking is a global shift. McKinsey’s Global Banking Annual Review 20262 estimates global banking net income at approximately $1.3 trillion in 2025. While the industry continues to operate at considerable scale, technology-led competition is changing expectations around responsiveness and efficiency.

The 2026 IIF–EY Global Annual Survey3, covering 65 financial institutions across eight regions, offers another perspective. It found that 98% of respondents reported measurable benefits from AI adoption, while approximately 70% indicated progress in moving agentic AI from development towards production.

These findings point to an important change in priorities. For product companies, this means making innovation easier to adopt, integrate and scale. Banks need capabilities that can evolve with their requirements without requiring a major technology transformation each time.

Building products that can adapt
A more workable approach is to build platforms that evolve alongside each institution. Modular architecture, configurable rules, open APIs and event-driven integration can help banks introduce capabilities progressively. Consider a missed loan repayment: can the system assess repayment history and account status, apply a collection strategy, distinguish a routine delay from a higher-risk situation and support the next step?

The same principle applies to transaction banking. Timely insight across payments, collections and cash positions can help institutions respond to liquidity requirements and operational exceptions. Intelligence depends on relevant information, controls and action. It is a product and architecture challenge, not simply an analytical layer.

Innovation also means protecting what matters
As banking platforms become more connected, cybersecurity becomes an even more important part of the product conversation. SBI’s October 21, 2026,4 stock-exchange disclosure highlights the importance of data protection. The bank said a preliminary analysis of approximately 375,000 records found 3,897 customer records containing personally identifiable information. It stated there was no material impact on core operations, financial performance or business continuity, and that an investigation was underway.

The financial consequences of security incidents are also significant. IBM’s 2026 Cost of a Data Breach Report put the average breach cost for India’s financial services sector at ₹40.9 crore, the highest among Indian industries covered. Security must inform architectural choices, data handling, identity management, API design, audit trails and recovery mechanisms from the outset.
For those developing banking technology, the implications go well beyond preventing unauthorised access. This reinforces why security and accountability must accompany speed and convenience.

Security must inform architecture, data handling, identity management, API design, audit trails and recovery from the outset. AI systems handling customer information or supporting financial decisions need clear permissions, monitoring and accountability. The RBI’s FREE-AI6 framework reinforces this balance between innovation, trust and responsible deployment.
The industry should consider innovation and security as connected priorities. In banking, a capability that improves speed or convenience but weakens control is not necessarily progress.

Agentic AI will test the maturity of product design
Agentic AI may coordinate activities across systems. In lending, an authorised agent could collect documents, consolidate information, prepare case summaries or flag exceptions, enabling employees to focus on judgement. But preparing a credit assessment should not confer authority to approve a loan. Decisions involving financial exposure or regulatory obligations require validation, controls and human intervention where necessary.

Moving AI from pilots to production is demanding; banking involves changing customer behaviour, incomplete information, exceptions and operational dependencies. Product teams must account for model performance, data quality, traceability, access controls and intervention mechanisms. Responsible autonomy must be designed in from the beginning.

Measure progress by the decisions technology enables
The value of intelligent banking should be assessed through business and customer outcomes, not the number of AI features a platform offers. In lending, that could mean improving turnaround time while maintaining underwriting quality, identifying repayment stress earlier or applying credit policies more consistently. In transaction banking, it could mean better cash visibility, fewer reconciliation exceptions or more accurate liquidity forecasts. Across both, the underlying systems must remain secure, reliable and adaptable as requirements change.

India has demonstrated how digital infrastructure can transform financial access at scale. The next opportunity is to make banking more responsive to customers and to the changing circumstances in which financial institutions operate. That calls for product development that looks beyond adding functionality and focuses on how systems interpret information, support decisions and adapt over time.

The practical learning is clear: intelligence, adaptability and trust cannot be treated as separate product priorities. They need to shape architecture, workflows, governance and the way new capabilities are introduced into live banking environments. Applying this thinking means building progressively, keeping human accountability where it matters and measuring innovation by the quality of decisions and outcomes it enables. As technology evolves, the enduring test will be whether banking products can be trusted to perform when it matters most.