"No company has yet solved how AI agents can effectively integrate legacy data to perform tasks efficiently. KT aims to solve this."
On the 3rd, Lee Sang-bong, Executive Vice President and Head of KT AX Data Lab, held a press study session at Club806 in Insadong, Jongno-gu, Seoul, announcing the establishment of 'Agetic Data Flow,' a company-wide data utilization framework. The initiative will be applied to 'Everyone's AI,' an agent designed for use by all citizens.
Recently, the number of companies utilizing AI agents for business operations has risen rapidly. However, as collaboration between AI agents increases, complexity in sharing data and operational standards also grows.
To address this, KT established 'Agetic Data Flow,' a data utilization framework enabling agents to accurately understand corporate data and apply it to tasks.
This framework consists of four stages: 'K Feature Map,' 'K Meta,' 'K Ontology,' and 'K Evaluation.'
K Feature Map is standardized data assets that AI agents can immediately utilize. It creates a unified data asset by standardizing and integrating enterprise data distributed across various systems, reducing the need to individually explore and process raw data.
K Meta is a data governance platform ensuring safe utilization of corporate data by AI. It enables AI agents to rapidly locate required data, understand data meaning and applicability, and grasp policies regarding personal and confidential information.
K Ontology is a semantic layer that allows AI to comprehend the enterprise's language within business contexts. It supports AI agents in understanding and executing internal terminology and operational manuals.
K Evaluation is an AI quality assessment system verifying the entire workflow of AI models and agents, including planning, tool invocation, and decision-making processes.
KT applied 'Agetic Data Flow' to improve its maturity across MyK, Sales Agent, and Small Business Owner Agent. It will also be implemented in K-Claw (tentative name), an upcoming business automation agent.
Additionally, KT plans to expand the scope of agents into diverse industries and business domains by integrating with the Everyone's AI service, while continuously advancing its Agetic AI ecosystem.
The executive stated, "Humans grasp business context and definitions through experience, meetings, and inquiries, whereas AI agents make decisions based on data. Drawing on accumulated assets, operational knowledge, and management experience in an environment with Korea's highest level of data complexity, we will demonstrate a new work operation model where humans and agents collaborate."
in TECH
