Generative AI startup Forty2Maru announced on the 12th that it participated in the February Capacity Building Academy's public AI special lecture held at the Korea Credit Guarantee Fund headquarters in Sejong City on the 11th and shared its public sector AI transition strategy and future vision.
At the special lecture, Lee Seung-hyun, Vice President of FortiTumaru, spoke on government innovation and digital transformation strategies in the AI era. He assessed the current state of public AX in South Korea and presented an implementation roadmap.
The vice president defined public AX as a process that goes beyond simply incorporating AI tools into existing operations and redesigns work methods and decision-making systems by embedding AI as the fundamental operating system of administration. He also cited structural limitations faced by the public sector, such as the gap between technological advancement and institutional change, rigid budget structures, and a shortage of specialized personnel, and suggested the need for flexible procurement methods and a phased investment strategy.
He also explained that it is important to prioritize defining administrative issues to be resolved rather than introducing AI technology itself, verify its effectiveness through small-scale demonstration cases, and then expand it throughout the organization.
The presentation also proposed AI innovations applicable to the work of local credit guarantee foundations. Key topics included the introduction of a review agent that automatically generates review and verification reports to reduce work time, a business support agent that searches for regulations and guidelines in real time, and a proactive agent that detects abnormalities, such as sharp declines in sales, to proactively respond.
In addition, FortiTumaru introduced six types of services applicable to public affairs, including question-and-answer using search augmentation technology RAG42 and artificial intelligence reading technology MRC42, natural language-based data analysis and visualization, report drafting support, policy planning support, 24-hour civil complaint consultation, and administrative process automation.
To mitigate the hallucination phenomenon identified as a limitation of ultra-large language models, Forti2Maru developed a technology that combines RAG42 and MRC42. It also offers a lightweight model, LLM42, specifically designed for specialized industries. It supports private mode for enterprises, enabling AI utilization in environments that protect internal data and sensitive information.
FortiTumaru plans to contribute to the transformation of public institutions into data- and intelligence-based digital platform institutions in the future.
- See more related articles
You must be logged in to post a comment.