The government will invest 18 billion won to develop core technologies for agentic AI that sets goals and makes judgments and takes action without human intervention.

The Ministry of Science and ICT and the Institute of Information & Communications Technology Planning & Evaluation (IITP) held a launch briefing for the "Agentic AI technology development for proactive action in the real world" project on Wednesday at the AI and Software Maestro Seoul Center.

The project is a key task under the government's recently announced "Agentic AI Initiative". It aims to link digital space and the physical real world to cut human involvement to 10 percent or less and secure technology that can autonomously plan and carry out complex tasks.

The ministry plans to develop core technologies to address limitations of existing AI agents, including interaction with the environment, tool use and generation, step-by-step plan revision and self-feedback, long-term memory and contextual understanding, multi-agent collaboration, and safety verification.

A total of 26 industry, academic and research organisations will take part, with NC AI, Medical Park, SimSimi and VENG as lead institutions. Key development tasks include an "agentic AI breast ultrasound" system that performs everything from image acquisition to analysis and drafting of readings, and an "agentic AI digital work colleague" that handles tasks by linking a company's enterprise resource planning system with email and calendars.

The project will also develop an "everyday companion and empathetic colleague" that understands a user's situation and provides emotional and daily-life support, as well as a multi-agent simulator based on the Model Context Protocol (MCP) that automatically performs physical analysis.

The government will provide a total of 18 billion won over 18 months starting this year. After a phased evaluation at the end of 2027, it will select 1 outstanding task and provide additional support for 1 year to develop products and services targeting the global market. The final output will be implemented as a prototype at technology readiness level 7 that operates in real environments and is commercially viable.

Jin-soo Lee (이진수), director general for AI Policy and Planning at the ministry, said, "It is important to secure core technologies that enable AI to make judgments and execute in real environments." He added, "We will provide support to strengthen the competitiveness of domestic agentic AI technologies and help lead innovation."