Fasoo, known as a security company, has changed its name to Fasoo AI. It is a sign of its intention to grow enterprise AI into a core business alongside security. Fasoo entered the AI business by launching its on-premises private LLM, 'Ellm', developed as an open-source model in 2024, and has been increasing references since last year.

It is focusing on combining document-related know-how built up through its document security business with AI technology to help companies use internal documents properly and securely. It is positioning its ability to provide an AI platform optimised not for a specific industry but for specific tasks within an industry as a point of differentiation.

Yoon Kyung-gu (윤경구), an executive director who oversees AI development as CTO of Fasoo AI, said companies need to closely combine work and AI to deliver results with AI. He said the company will unveil the latest technologies supporting that at its annual customer conference, FDI Symposium 2026, to be held on April 15.

According to Yoon, Fasoo AI's customised AI platform strategy can be summed up as combining RAG (Retrieval-Augmented Generation) and AI agent technology based on Ellm, a small language model (sLLM).

Yoon said, "Ellm is not a simple LLM but an enterprise AI platform." He said 80 percent of the AI capabilities companies need are implemented in advance in the platform, while the remaining 20 percent is provided optimised for each customer's environment. He said companies find it difficult to use AI properly unless it is implemented as specific applications at the department level, and that even within the same domain, each department's needs differ. He repeatedly stressed that optimisation is important in AI.

RAG helps companies use internal data more effectively by using LLMs, but there has also been no shortage of comments that using it properly requires attention to many details. In this regard, Fasoo AI is focusing on knowledge graphs and AI agents.

A knowledge graph is a database that organises facts into clear relationships and is useful for reducing LLM hallucinations and raising the level of inference. Yoon said, "To get results from RAG, there is a need to build a knowledge graph in advance. We will disclose our achievements on this at the FDI conference."

On AI agents, he said, "RAG is heavily affected by the characteristics of the data itself, but AI agents can raise satisfaction for specific uses." He said the company is focusing on developing technology to improve accuracy in a way that agents debate based on web searches and internal search results.

Fasoo AI said there is also strong synergy between its security speciality and its AI business. Yoon stressed, "The RAG Fasoo AI provides is not a simple RAG but RAG that understands documents well," adding, "By applying security policies to vector database searches as well, it can be used without concerns about document security."

Fasoo AI has focused since the launch of Ellm on supporting information retrieval-based work across the public sector and companies, but it has recently been expanding its application areas. Yoon said, "In addition to information retrieval, various requirements are coming in, such as sales outlook analysis."