After the technology to create, the next competition will be decided by control infrastructure

Original: IT Chosun — Agentic AI Map 2026 · Genesis Lab, from AI Agent Design to Control and Execution… Targeting Enterprise 'Agent OS'

Genesis Lab was featured in 'Agentic AI Map 2026', a project co-created by IT Chosun and the National Information Society Agency (NIA). This series examines the domestic AI agent industry by company, and the completed map will be unveiled at the 2026 Korea AI Awards.

The article describes Genesis Lab as a company that helps design, deploy, and control agents. In Agentria, business practitioners can build agents by connecting nodes on a visual canvas. If complex logic is required, it can be extended using Python. Design, testing, deployment, and operational management are connected within a single environment, and once a workflow is created, it can be reused as a component for other agents.

AI results can vary each time they are executed. Therefore, a feature has been included to log the work process and outcomes, and rerun them under the same conditions. This is a mechanism designed to trace where errors occur and operate agents within the organization's permissions and security policies. The company explains this structure, which also bundles APIs and Model Context Protocol (MCP) deployment management, as an 'Agent OS'.

ViewinterHR analyzes both verbal and non-verbal information from interview videos to be used in recruitment and competency evaluation. It is used by over 300 companies and, in 2024, received the HR field AI reliability certification from the Telecommunications Technology Association (TTA). The evaluation criteria accumulated in recruitment fields, outcome explanations, and bias management experience have led to the design of agent governance.

Agentria recorded approximately 2.1 billion KRW in sales in its first year of launch and secured 15 white-label sites. Cumulative investment stands at approximately 20 billion KRW, with Series B Bridge investment currently underway. As next challenges, the company pointed to an ontology layer that structures in-house knowledge and a governance layer that controls execution. A beta version of the Agent OS, which processes natural language requests in accordance with authorization, is also in preparation.

CEO Young-bok Lee said, "The technology to build agents has already become commonplace." He explained that the game will be decided by the infrastructure that securely controls everything from ontology to governance and orchestration.