Samsung SDS unveiled its latest agentic AI offerings, with a view to bring its agents to South Korea’s public and financial services.

The upgrades are to its existing AI offerings of FabriX, Brity Copilot, and Brity Automation, giving them higher autonomy to remove the amount of steps and prompts required by previous iterations. This could possibly remove demand on network bandwidth, if balanced with agentic AI's inherent impact on network capacity.

FabriX is an enterprise-grade generative AI (genAI) platform that includes Agent Studio, enabling companies to build custom AI agents tailored to tasks, departments, or roles. It is offered along with Brity Works Suite, Samsung SDS’s integrated enterprise collaboration platform, which features collab tool Brity Copilot, and RPA-platform Brity Automation.

Those who follow agentic AI may be confused as to how a copilot can be an agent; likewise for an RPA tool. The confusion would be understandable, especially from those tired of agentic hype.

Either way, the upgrade puts SDS another step forward in AI ahead of local cloud rivals such as Naver Cloud and LG CNS, who are yet to offer agentic AI solutions (for better or worse).

As part of its targeting of public services, Samsung SDS is planning to offer FabriX as a software-as-a-service (SaaS) platform for the South Korean government's electronic document and task management system. This environment would see agents dealing with sensitive data, thus necessitating stricter security controls on the network side.

"By integrating agentic AI capabilities into Brity Copilot and Brity Automation, we aim to help businesses accelerate automation," Samsung SDS VP Song Hae-goo noted in a statement. "After solidifying our domestic presence this year, we plan to expand into the European market next year."

That may be an ambitious plan as Samsung SDS has yet to make a dent in the European cloud market. That market will also pose stricter data residency rules as the clamor for European data sovereignty increases.

Another possible hurdle on the horizon is that 40% of agentic AI projects may be canned by 2027, according to research firm Gartner.