YES Bank's Chief Information Officer Mahesh Ramamoorthy has outlined the bank's strategic approach to artificial intelligence implementation, emphasizing the potential of homegrown AI models for enhancing operational efficiency. According to reports from The Hindu BusinessLine, Ramamoorthy highlighted that AI is taking centre stage from a strategy perspective, with the bank actively progressing AI into areas that will be meaningfully important for their operations. AI is central to our strategy and the bank is actively working to integrate AI into operations that will deliver meaningful value.
The bank is currently addressing tokenization cost challenges as a primary focus for the CIO department. As reported by The Hindu BusinessLine, Ramamoorthy explained that the biggest cost of AI lies in how tokens are used and where they are deployed. The bank is examining how to restructure their ecosystem to derive maximum benefits by using the right model against specific use cases, with local models like Sarvam or BharatGen potentially solving document or speech-to-text challenges more effectively due to their localization and Indianization. LLMs will start becoming more commodity as the technology matures, but the bank must orchestrate requests effectively to optimize costs.
Ramamoorthy confirmed that YES Bank is open to engaging with more homegrown models as long as they derive value and have appropriate guardrails. According to The Hindu BusinessLine, he emphasized that maturity is starting to emerge in these localized solutions, which should eventually lead to better efficiencies. The bank is particularly interested in using local models for document verification, extraction, and other localized tasks, rather than frontier LLMs that are 10 times the price for similar capabilities. There is no harm as long as you have the right guardrails - knowing what data is getting passed and ensuring it doesn't compromise security.
Regarding AI's impact on technology debt, Ramamoorthy indicated that it will take some time before AI helps rationalize existing tech debt. As reported by The Hindu BusinessLine, he expects the ability to use AI to modernize the ecosystem and derive AI benefits to be 12-18 months away. He noted that bank technology implementation requires careful consideration, as bringing something new to the banking sector is not like a snap of a finger, requiring careful planning and execution. The ability to use AI to modernise your ecosystem or to use the benefits of AI is still about 12-18 months away.
According to The Hindu BusinessLine, banks are beginning to take AI into the frontline, directly exposing it to customers in assisted formats. Ramamoorthy explained that banks are regulated entities first and foremost, with trust being the fundamental aspect. Currently, AI capabilities are focused on operational efficiency, cost efficiency, and risk management. As banks mature, AI-led capabilities are expected to expand into model governance, with the RBI having issued expectations in this area. Banks will start taking AI into the frontline, which is direct customer - it may be assisted but directly exposing it to customer will take its course.
