South Korea's financial sector is transitioning from experimenting with artificial intelligence (AI) to applying it in real-world operations. As AI rises to the forefront of executive agendas, financial institutions are accelerating their AI transformation (AX) and moving beyond mere implementation to actual customer service applications.
On August 27, AWS Korea held a press conference at the Seoul Chamber of Commerce during the 'AWS Financial Services Forum 2026.' Attendees included Choi Gi-young, head of Anthropic Korea, and Noh Kyung-hoon, head of AWS Korea's financial services and public sector.
Choi emphasized, "The most significant change in AI adoption in finance is that it has become a priority for top executives." He noted that as interest from CEOs grows, there is a rapid shift from pilot programs and proof of concepts (PoCs) to actual implementation in business operations.
Choi added, "Some CEOs are even taking coding training themselves," highlighting the increasing prevalence of top-down AI adoption in the country.
Noh stated that as AI moves into production, it is beginning to have a substantial impact on business. He remarked, "In the past two to three years, we were in the stage of exploring what AI could do and conducting PoCs. This year, however, there has been a significant increase in questions about how to operate and utilize agents."
Shinhan Card has sequentially launched 15 customer-facing AI agents for tasks such as card recommendations and credit assessments for foreigners. Kakao Insurance is also applying AI to enhance productivity for developers and operational staff.
The criteria for evaluating AI are also changing. Initially, the number of agents created was a key metric, but now the focus has shifted to the impact on business, such as reductions in work hours and costs. Noh noted, "We are seeing cases where tasks that used to take days are now completed in hours, and those that took hours are now done in minutes."
However, as AI usage expands in the heavily regulated financial sector, establishing control systems to ensure security and reliability has become a critical challenge. Financial institutions must ensure not only the performance of models but also their accuracy, consistency, and explainability, necessitating a secure and controllable environment.
Choi stressed, "In the financial sector, it is essential that AI can provide consistent and accurate answers, even when asked repeatedly, and that it can explain why it provided those answers." He added that as agents connect to multiple systems, the importance of safety measures, or guardrails, increases.
AWS and Anthropic are enhancing their partnership to meet these demands from the financial sector. Anthropic provides frontier AI models and safety technologies, while AWS supports permission management, monitoring, and security and control environments through Amazon Bedrock and AgentCore.
Noh stated, "We need to be able to examine what data AI uses, what results it produces, and how those results are generated," emphasizing that AI and agent platforms must be built in an auditable manner.
In this context, Anthropic is preparing to offer its latest Claude model in the AWS Seoul region. Choi noted, "If we can process inference within the Seoul region, it will reduce concerns about data leaving the country while allowing the use of the latest frontier models."
AWS and Anthropic have proposed 'AI Native Finance' as the next step for financial AI, which involves redesigning work and organizations around AI.
Noh expressed optimism, stating, "I expect to see the form of 'AI Native Finance' emerge in a year. If AI is handling these tasks, we will need to rethink how to change workflows, leading to the emergence of 'AX Everywhere.'"
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.