Artificial intelligence (AI) agents represent the next inflexion point in banking automation, shifting the paradigm from rule-based orchestration to cognition-driven execution. In an interaction with TNIE’s Padmini Dhruvaraj, Infosys Finacle CEO Sajit Vijayakumar tells AI agents will make traditional core banking and batch-based payment systems obsolete, forcing banks to overhaul their technology stacks for real-time and autonomous operations. Edited excerpts:
AI agents are collapsing timelines for complex banking and IT work. How exposed is the core banking and payments software model to this shift? What is Finacle doing?
AI agents represent the next inflexion point in banking automation, shifting the paradigm from rule-based orchestration to cognition-driven execution. Powered by advances in large language models (LLMs) and GPT-class reasoning systems, these agents can interpret intent, synthesise context across fragmented data landscapes, and autonomously coordinate multi-step financial workflows. Unlike earlier automation layers that optimised discrete tasks, AI agents compress decision cycles across key business processes and functional workflows. The result is a structural collapse in timelines for complex banking processes, enabling near-real-time responsiveness in domains historically constrained by manual review, siloed systems, and sequential processing logic.
Traditional monolithic cores and batch-oriented payment infrastructures are structurally misaligned with agent-driven, real-time orchestration models. Future-ready platforms will need composable architectures, event-native processing, and policy-aware control layers that allow AI agents to operate safely within deterministic guardrails. Finacle offers a composable digital banking platform, built on the foundations of a 100% open architecture - it exposes core banking and payments services as modular, MCP‑compatible tools, enabling banks to set up agentic‑driven use cases where AI can securely discover, invoke, and operate governed banking functions. Finacle AI’s offerings include both predictive and generative AI assistants that support a host of use cases across lines of business and customer/user journeys. Finacle has also built AI assistants to augment key SDLC processes and introduced agentic AI capabilities to facilitate automated user interface development.
Where does Finacle create defensible value that banks cannot replicate using in-house AI or third-party agentic platforms?
Finacle’s defensible value in the era of AI stems less from standalone algorithms and more from the institutional depth of its enterprise data and domain architecture. Sustainable AI advantage in banking is fundamentally a data problem - requiring governed, high-fidelity, and semantically consistent information at scale. Finacle addresses this through an enterprise-grade data platform that combines industrialised data engineering pipelines, governed lakehouse constructs, and banking-aligned canonical models inspired by standards such as BIAN. This foundation materially reduces the time, risk, and fragmentation banks typically face when attempting to operationalise AI independently, transforming raw data availability into AI-ready intelligence. Layered on this foundation is the Finacle AI Platform - a unified, no-code environment that industrialises the full lifecycle of explainable AI across development, deployment, monitoring, and continuous optimisation. Finacle’s investments in domain-tuned generative and small language models, combined with enterprise-wide Responsible AI frameworks from Infosys Topaz, further extend this moat. This convergence defines Finacle’s differentiation: translating AI potential into production-grade, regulator-ready, and economically sustainable banking intelligence.
Digital payments are a critical infrastructure but hard to monetise. What revenue models are actually working for banks today?
Despite banks funding the entire security, scale, and stability of the country’s payment system, the real economic value is increasingly shifting toward the ecosystem surrounding the payment rather than the payment event itself. Credit originated on payment flows, including credit-on-UPI and embedded merchant or consumer lending, is emerging as the most scalable lever given structurally stronger margins and superior behavioural risk insight. Merchant-facing value-added services - such as analytics, reconciliation, loyalty enablement, and working-capital solutions - enable banks to monetise participation across the payment journey rather than the transaction alone. In parallel, API monetisation and Banking-as-a-Service models are generating recurring ecosystem income as banks power partner distribution, while sustained digital payment adoption continues to deepen granular, low-cost deposits and float balances, an often underestimated but meaningful contributor to profitability. Realising this transition requires a modern, resilient payment architecture. Cloud-native, ISO 20022-based payment hubs, such as Finacle’s, can enable banks to participate actively across multiple schemes at population scale, generate rich transactional intelligence and facilitate nuanced financial services without compromising resilience or compliance.
Is AI in banking delivering measurable profit uplift, or is it still largely a cost and compliance investment?
Across much of the industry, AI investments continue to concentrate on internal automation, regulatory reporting, risk surveillance, and workflow optimisation. These applications materially strengthen governance, resilience, and operating cost structures, but their contribution to top-line growth is indirect. Scaled deployments in personalisation, intelligent credit, fraud signal detection, and sales augmentation are beginning to produce quantifiable shifts in efficiency ratios, loss containment, and revenue conversion. These outcomes signal an early but credible transition from productivity tooling to earnings leverage.
Five years from now, which part of today’s banking technology stack do you believe will become irrelevant?
The components of the banking technology stack at the highest risk of irrelevance today are static channel applications, tightly coupled middleware, and batch-oriented integration patterns - designed for predictability rather than adaptability - that are already becoming constraints in an era defined by real-time expectations, ecosystem connectivity, and AI-driven decisioning. Equally vulnerable are rule-heavy, manually governed operational processes that sit outside intelligent automation. Traditional workflow engines, fragmented data silos, and opaque decision logic are increasingly incompatible with a future where AI copilots, autonomous orchestration, and real-time risk sensing become embedded in everyday banking operations. The stack will not disappear overnight, but value will migrate rapidly toward modular, cloud-native, API-first, and AI-augmented architectures that allow banks to change faster than the market around them. What will remain durable is not a specific technology category, but a design philosophy: composability, resilience, explainability, and continuous evolution.