"Reduced GPU Usage by 70%"... BC Card Unveils Agentic Financial AI in the US

On May 14 (local time), BC Card announced on the 18th that it unveiled the case of its "Agentic AI Operation Platform" for the first time on the global stage at the "Red Hat Summit 2026," the world's largest open-source technology conference held in Atlanta, USA.

On the 14th (local time), Tae Young Lee, Head of the AX Team at BC Card, presented a case study on the application of the Agentic AI operational platform service at the Red Hat Summit 2026 held in Atlanta, USA. BC Card

The "Red Hat Summit" is a global event organized annually by the global open-source company Red Hat, having first launched in New Orleans, USA, in 2005. The open-source, cloud, and AI technology conference attracts tens of thousands of IT leaders, developers, and engineers from around the world every year.

Agentic AI refers to artificial intelligence that goes beyond simple responses to make autonomous judgments and take action. The "Agentic AI Operation Platform" utilizing this technology clusters and processes multiple small language models (SLMs), each specialized by domain, simultaneously, instead of relying solely on a single large language model (LLM) for all work as in the previous approach.

While LLMs have limitations such as high operating costs and slow processing speeds, SLMs are optimized for specific tasks, enabling not only stable outcomes but also improving inference speed by up to three times and reducing graphic processing unit (GPU) resource usage by up to 70%.

At the event, BC Card presented a case study on efficiently distributing expensive GPU resources by leveraging the know-how accumulated from processing over 10 billion payment transactions annually with AI technology. Based on these technological achievements, the company also introduced currently operating services that go beyond simple AI adoption, including: ▲Eat.pl (real-time analysis of consumption patterns and restaurant recommendations), ▲BCGPT (supporting employee work through a finance-specialized generative AI platform), and ▲MOAI (connecting multiple AIs to implement automated work processes), drawing attention from global enterprises.

Alongside R&D and technology development, the company is also pursuing various activities to contribute to the advancement of global AI infrastructure. BC Card has previously released approximately 38 LLMs developed in-house, along with Korea finance-specialized datasets, as open source on "Hugging Face," a global LLM open-source hub, making them available to developers worldwide.

Sungsoo Oh, Head of the AI Data Division at BC Card, stated, "Our presentation at this Red Hat Summit was a meaningful occasion where BC Card's global-level agentic AI engineering capabilities were officially recognized on the world stage," adding, "Representing the Korean financial sector, we will continue to demonstrate our identity as a technology leader contributing to the open-source ecosystem, going beyond simple AI adoption cases."

This content was produced with the assistance of AI translation services.