The scene at the AWS Financial Services Forum Seoul 2026, held on Aug. 27 at the Korea Chamber of Commerce and Industry, where Amazon Web Services Korea shared financial-sector AX trends and recent cases.

Key messages included that financial companies are speeding up efforts to put AI into production beyond proof-of-concept testing, and that executives play a major role in financial-sector AX.

Kyung-hoon Noh (노경훈), head of financial business at AWS Korea, took the stage for a keynote speech and a media briefing and stressed that cases of AI agents being deployed in production are increasing in the financial sector from this year.

“Cases are increasing where companies move from PoC to actual application,” he said. “Because it directly affects business, not only IT departments but many departments are paying attention to AI.” He said Shinhan Card has been sequentially releasing 15 customer-facing agents from this year. He said functions such as credit card recommendations and credit card screening for foreigners can already be used in production.

He said that across all areas of finance, including banking, cards, securities, life insurance and fintech, companies previously were largely testing what they could do with AI, but the nature of questions has changed this year. He said questions have surged on how to operate agents and use them safely, and how to monitor and run them so that agents created at scale across departments do not become duplicated or trapped in silos.

Ki-young Choi (최기영), head of Anthropic Korea, who attended the keynote and media briefing with Noh, said that as with other industries, financial companies are also thinking hard about ways to apply AI to actual work rather than staying at the pilot or PoC stage. He said AI is already being widely used, particularly in asset management or product design areas that do not directly handle customer data.

Security and safety issues and data reliability were cited as factors the financial sector should pay attention to when deploying AI in production.

Choi said security and safety are extremely important in South Korea because they are tied to compliance with financial regulators' rules. He said that is why Anthropic implements safe AI by applying the principles of “Constitutional AI”. He said AI must be able to explain the basis for its conclusions, and how to ensure security is among what is often demanded at the production deployment stage. He added that companies also need evaluation criteria to judge whether AI is executed in line with business objectives and leads to results.

Noh highlighted that it is also important how well data is prepared for an AI environment.

He said expectations for AI have risen sharply over the past 2 to 3 years, but when IT departments try to create environments to build and use AI agents, data is often not prepared at all. He said ontology building and “AI-ready data” projects are therefore being actively pursued.

Another notable point is that executives account for a large share of AX. Choi said he sensed that executives in the financial sector show strong interest in AX, to a degree that personally surprised him.

“AI has become an agenda emphasized by top management,” he said. “CEO-level executives show interest to the extent of receiving vibe coding training themselves. Executive interest in AI is far greater than when they pushed cloud or digital transformation in the past.”

Noh also cited Shinhan Card, saying that at chairman level it has emphasized agents for several years. He added that demands related to AI are pouring in within organizations and that financial companies' IT departments must also create environments where AI can be used safely and stably. He said strong executive will is important to persuade people what is needed while managing internal expectations.

Woo-young Kim (김우영), head of the AI strategy office at Hyundai Capital, also shared the company's AX experience at the event, putting weight on the role of executives.

“In AX, executive interest is important. Companywide AI transformation cannot be done just because a specific organization works hard,” he said. He said Hyundai Capital began full-scale AI transformation work from early last year and holds a monthly AI-only meeting with all business leaders, including the president, participating. He said the meeting shares AI technology trends and intensely debates whether AI priorities are aligned in direction. He said decisions made top-down with full executive support enabled a multi-year AI transformation master plan and detailed execution plans to proceed stably.

Kim also stressed the importance of directly hearing employees' views, as much as executive interest.

According to Kim, Hyundai Capital ran the effort in an executive-led way for 1 to 2 years and began to wonder whether it was sufficiently hearing the voices on the ground. He said around that time AWS proposed a vibe coding hackathon, which led to holding one.

Kim said the most important principle for the hackathon was to make it an event centered on general employees rather than developers.

He said developers were not allowed to form teams and only participated as supporters. He said the company held an orientation and workshops before applications to raise interest, and AWS supported training in that process. He said the president also attended from the orientation and delivered a message that the event was important for the company.

Kim said the response to the hackathon and the outcomes exceeded expectations. He said he initially expected about 10 tasks to advance to the final round, but far more teams applied. He said many submitted ideas were complete enough to reach the final round, and several teams among them carried out actual hackathon projects. He said the company had planned to develop only some award-winning tasks to production level, but because the outputs were highly complete, it decided to reflect most of the tasks that reached the final round in next year's business plan.