For India’s banking sector, the AI opportunity is no longer a question of whether to adopt the technology, but where, how and how fast to deploy it. As banks move from experimentation towards scaled adoption, the next phase of AI transformation will be shaped as much by trust, governance and economics as by the technology itself.
Speaking on the sidelines of Yes Bank’s AI Fest 2026, Mahesh Ramamoorthy, Chief Information Officer, Yes Bank, said financial institutions recognise that AI is set to become an inevitable part of the industry’s future. However, the highly regulated nature of banking means institutions cannot simply rush to deploy AI across customer-facing processes.
“The primary goal is trust,” Ramamoorthy said, pointing out that AI, unlike traditional deterministic systems, is inherently probabilistic. For banks, the challenge is therefore to identify the right use cases and establish the right guardrails before accelerating adoption.
This is already influencing where AI is gaining traction. Much of the early momentum is emerging in areas where AI can augment employees, improve decision-making and increase operational efficiency without independently making critical customer decisions. Contact centres, post-call analysis, customer query understanding, sentiment analysis, fraud investigations, enhanced due diligence and the preparation of credit appraisal memos are among the areas where AI use cases are progressing rapidly.
The model emerging in banking is not one of complete autonomy, at least for now. Instead, Ramamoorthy sees a “human in the final loop” approach becoming increasingly important, particularly for decisions where trust, accountability and regulatory compliance are critical.
This cautious approach reflects the broader maturity curve of AI adoption across the financial services sector. While some banks have moved further ahead in deploying AI, others are still catching up, and a third group is working through where and how the technology can create meaningful value. Over the next 18 to 24 months, these differences could become more pronounced as institutions move from pilots to production-scale deployments.
One of the biggest factors determining this transition will be the economics of AI.
Today, the cost of deploying large language models remains a significant consideration. Ramamoorthy noted that using massive models with billions or trillions of parameters for narrow enterprise tasks may not always make economic sense. The industry, he believes, will increasingly move towards choosing models based on the specific use case rather than assuming that the largest model is always the best solution.
This could open the door for smaller, specialised and locally trained models, particularly for applications involving Indian languages, speech-to-text and other contextual requirements. The objective will increasingly be to achieve the right balance between performance, cost and scale.
“Where you deploy which model will become important,” is the underlying message. AI adoption in banking will not be about selecting a single technology platform, but about building an architecture that allows institutions to match the right model to the right business problem.
This philosophy was also reflected in Yes Bank’s AI Fest, held at its Mumbai headquarters on August 27, 2026. The initiative brought together more than 29 technology partners, alongside teams from across the Bank, including business, product, risk and compliance, to explore AI use cases and scalable implementation strategies.
Built around the theme of “Intelligence, Innovation and Impact”, the event was designed to move conversations beyond AI experimentation and towards identifying solutions mature enough for real-world adoption. By bringing technology providers and multiple internal stakeholders together, the bank sought to create a more collaborative approach to AI innovation.
Ramamoorthy said the initiative aligns with Yes Bank’s broader strategy of building AI capabilities methodically, supported by governance frameworks and focused on sustainable business impact.
That governance-first approach is particularly significant in financial services. Yes Bank has established a Board-approved AI Governance Policy, an AI Council and an AI Centre of Excellence to oversee AI implementation and the model lifecycle. The bank has also introduced Responsible AI training as it expands AI deployment across select processes and customer journeys.
The larger message from the AI Fest and Ramamoorthy’s perspective is clear: banking’s AI transformation will not be won simply by being the first to deploy the most advanced model. The winners may instead be those that can combine innovation with trust, governance and a clear understanding of business value.
As AI costs evolve and specialised models become more accessible, banks are likely to accelerate adoption. But the transition to autonomous AI will remain measured. For now, the sector’s focus is on building confidence in the technology—using AI to make employees more productive, decisions more informed and operations more intelligent, while keeping human oversight firmly in place where it matters most.
For banking, AI’s next chapter may therefore be less about replacing humans and more about creating a trusted intelligence layer around them. And as institutions move from isolated pilots to enterprise-wide adoption, the ability to deploy the right AI, for the right use case, at the right cost could become the real competitive advantage.