Korea Agency for Infrastructure Technology Advancement Wins 'Innovation Award' at Nationwide Public Institution AI Challenge

Automatically generates a 150-page draft from one page of project information
Six AI agents work together, cutting planning time by 31.8%
Developed in-house without outsourcing, with plans to expand it as a government standard model

The Korea Agency for Infrastructure Technology Advancement, a research and development (R&D) institution under the Ministry of Land, Infrastructure and Transport, developed an automated system for generating national R&D planning reports using artificial intelligence (AI) and won the AI Innovation Award in an evaluation for public institutions.

Officials from the award-winning institutions, including the Korea Agency for Infrastructure Technology Advancement, pose for a commemorative photo at the 2026 Public Institution AI Innovation Challenge awards ceremony held at Nuritkum Square in Seoul on the 28th. Korea Agency for Infrastructure Technology Advancement

The Korea Agency for Infrastructure Technology Advancement said on the 28th that it had won the AI Innovation Award in the Demonstration and Proof of Concept (PoC) category at the 2026 Public Institution AI Innovation Challenge, held at Nuritkum Square in Mapo-gu, Seoul. Hosted by the Ministry of Finance and Economy and the Ministry of Science and ICT, and organized by the National Information Society Agency, the competition was launched to identify examples of AI-driven public service innovation that people can experience firsthand. After evaluating public institutions nationwide, the organizers selected 30 projects for the final round.

The AI-powered automated national R&D planning report service submitted by the Korea Agency for Infrastructure Technology Advancement brings people and AI together as a team to plan R&D projects. Its core operating principle is a team of "six AI research assistants," each specialized in a different field. Previously, many experts had to spend months finding information themselves and meeting to develop an initial blueprint for a project. With this system, however, six AIs divide up the work like staff members.

By entering only one page of key project information, users can have multiple AI agents work together to draft a planning report of around 150 pages. The system also uses customized memory technology that carries context from one section to the next, maintaining the logical flow and consistency of extensive documents during generation.

The AI does not generate an entire document on its own. Instead, the system uses a "human-AI teamwork" approach in which the expert responsible for each stage of drafting reads, revises, or approves the AI-generated content. The system replaces the existing approach, which relied on a small number of experts, with data analysis while adding human insight and intuition to preserve expertise.

The system was used in planning projects including the Hyper Tube project, a next-generation high-speed railway, and its work was reviewed by eight experts from various fields. As a result, it cut the time needed for planning by 31.8% compared with the usual year-long process. This collaborative approach has also been featured in a journal indexed in the Korea Citation Index (KCI).

The system is considered a low-cost, high-efficiency innovation developed from ideas contributed by the institution's own employees, without expensive external consulting or complex commercial software. It is easy for other institutions to adopt: their employees can start using this "AI colleague" right away by simply exchanging a file, without building costly server infrastructure. The Korea Agency for Infrastructure Technology Advancement is currently using the system as a reliable assistant for planning new projects in construction, railways, aviation, and other fields.

"This is the result of our employees designing for themselves how to work with AI and turning that approach into tangible results," said Kim Jeonghee, president of the Korea Agency for Infrastructure Technology Advancement. "We will expand its use to every department within the agency and promote it as a standard planning model for government R&D, so that taxpayers' money can be invested more quickly and accurately in technologies that truly need it."

This content was produced with the assistance of AI translation services.