At the International Conference on Sustainable Computing and Communication Technologies (ICSCCT 2026), organized by the University of Malta’s Faculty of Information and Communication Technologies and published in partnership with Springer, Subhankar Panda took to the stage as a keynote speaker to make a case that few in the audience discussed, but many are still slow to act on: the next decade of banking will be absorbed by its central planning and not just learning. automatic back-office tools.Subhankar’s talk moved beyond the familiar narrative of AI as a cost-cutting layer built into legacy systems. Instead, he framed AI and ML as tools that should sit within a bank’s financial planning function, shaping how risk is priced, how liquidity is forecasted, and how decisions that once took committee weeks are now taken in model minutes.“Banks that treat AI as an add-on will continue to optimize yesterday’s processes. Banks that treat it as a planning partner will start asking better questions about tomorrow’s balance sheet.”From automation to anticipationA recurring thread in the address was the shift from AI that automates known tasks to AI that anticipates unknown ones. Subhankar pointed to machine learning models that are now able to stress-test portfolios against scenarios that human analysts would rarely think of building on their own, and to pick up early signals in transaction data long before it appears in a quarterly report.He argued that this shift changes the financial planner’s job as much as the engineer’s: less time spent crunching numbers, more time spent deciding what the numbers mean.“The value is not in the model producing an answer. The value is in the banker knowing what question was worth asking.”Trust, governance and limits of the modelSubhankar was careful not to present the adoption of AI in banking as a purely technical issue. He dedicated part of the presentation to governance: the need to explain oneself in credit and risk decisions, the regulatory weight that financial institutions have and the reputational cost of deploying models that cannot explain their own reasoning. In an industry where a single miscalibrated model can impact customer confidence and compliance exposure, he suggested that responsible deployment matters as much as capability.This emphasis connects to the larger argument underlying Panda’s recent work: that reliability engineering and AI adoption are not separate conversations. His writing on AI-driven test automation in enterprise delivery has made a relatable point in the software world: as systems grow more autonomous, the discipline of verifying them must grow just as fast, or the speed AI promises becomes a risk rather than an advantage. Applied to banking, the same logic holds: an AI-based planning system is only as reliable as the evidence and governance built around it.A call for institutional patiencePanda closed by warning against treating AI transformation in banking as a one-off project with an end date. Instead, he described it as an ongoing capability that must be funded, staffed and continually reviewed, more closely aligned with how institutions approach risk management than how they approach software deployment.“The banks that do well in five years are the ones that treat it now as infrastructure, not the ones waiting for a finished product to buy.”ICSCCT 2026 attracted researchers and practitioners from the fields of informatics, sustainability and applied technology for two days of sessions at the University of Malta, with proceedings to be published via Springer.