Agentic Artificial Intelligence is coming of age and opening new opportunities across financial services. Banks are increasingly using agentic Artificial Intelligence to optimize processes, navigate complex systems, and sift through large volumes of unstructured data to make decisions and take actions either with or without human intervention. Sameer Gupta, Americas financial services Artificial Intelligence leader at EY, notes that the maturing of agentic Artificial Intelligence makes large scale process automation possible in ways that rules based approaches like robotic process automation could not, and that this shift affects cost, efficiency, and customer experience.
Practical implementations span customer service and back office functions. The article cites examples such as responding to customer service requests, automating loan approvals, adjusting bill payments to align with regular paychecks, and extracting key terms and conditions from financial agreements. Firms view these capabilities as transformative for customer experience and operational models. Murli Buluswar, head of US personal banking analytics at Citi, says that a company’s ability to adopt new technical capabilities and rearchitect how the firm operates will determine which firms succeed and which get left behind, and that people and organizations must recognize that work will look meaningfully different.
Adoption is already measurable. A 2025 survey of 250 banking executives by MIT Technology Review Insights found that 70 percent of leaders say their firm uses agentic Artificial Intelligence to some degree, with 16 percent reporting existing deployments and 52 percent running pilot projects. More than half of respondents said agentic Artificial Intelligence systems are highly capable of improving fraud detection, at 56 percent, and security, at 51 percent. Other prominent use cases included reducing cost and increasing efficiency, and improving the customer experience, each at 41 percent. The piece also notes that the content was produced by Insights, the custom content arm of MIT Technology Review, and was researched, designed, and written by humans, with any Artificial Intelligence tools used limited to secondary production processes under human review.
