Defense researchers in South Korea can now use a homegrown generative AI platform for classified work, eliminating the need to rely on commercial tools.
Developed by the Agency for Defense Development (ADD), Add+i runs entirely on the agency’s internal network to search internal regulations, summarize documents, translate text, and streamline a range of research tasks.
It was built using open-source AI models, giving ADD greater control over customization, security, and deployment.
From ‘Vibe Coding’ to Faster Research
A key feature of Add+i is support for “vibe coding,” allowing researchers to turn natural-language prompts into working software while accelerating code development, debugging, and testing.
ADD said the tech has already been used to develop a geospatial information management system that integrates spatial data across the organization.
It has also reportedly reduced the time spent searching internal information and handling repetitive administrative work, giving researchers more time to focus on higher-value research and development.
Agentic AI Up Next
ADD is already developing Add+i 2.0, an agentic version designed to understand intent, connect with internal tools and systems, and carry out tasks more autonomously.
The upgrade will also support the Model Context Protocol, enabling AI agents to interact with internal defense systems and data sources.
ADD Director General Lee Geon-wan described Add+i as the agency’s first major step toward integrating generative AI into everyday defense research.
“Add+i is the first challenge for our institute’s employees to utilize generative AI in their actual work,” Lee said.
“Through this, we plan to continuously expand our R&D capabilities by connecting the data accumulated by the institute and the knowledge of our staff to the AI ecosystem.”
