Ourbox Accelerates Logistics AI Transformation with Proprietary LLM

South Korean AI firm Ourbox is spearheading a logistics-specific AI transformation (AX), leveraging its proprietary Large Language Model (LLM) and real-world operational data. The company aims to overcome the complexities of logistics environments that have challenged the application of general-purpose AI, thereby enhancing operational intelligence and efficiency.

Ourbox's self-developed Korean LLM, 'Ourbox-31B-JGOS,' achieved a notable second place in the K-AI Leaderboard's over-30B parameter category, scoring 0.612. The model received high marks for its comprehension of Korean context, specialized knowledge, and complex reasoning abilities. This development is particularly significant as logistics centers often deal with intricate, real-time variations in order fulfillment, inventory, and customer-specific handling procedures, which standard AI struggles to address effectively.

By combining its validated Korean language understanding and reasoning skills with extensive logistics data and operational expertise, Ourbox has developed a "logistics-specialized AI." Initially, this AI will be integrated with Ourbox's 'MATE' logistics system to support on-site tasks such as searching product and order information, assisting workers, and automating documentation. Future applications include expanding into operational decision support, such as predicting workloads, identifying inventory discrepancies, and recommending task priorities.

“

This accomplishment lays the groundwork for logistics companies to secure their own AI technology capabilities and solve on-site problems.

Furthermore, Ourbox is adopting on-device AI to improve field applicability and reduce costs. In collaboration with AI tech firm Bidraft, they are deploying the CPU-based 'POCKET' platform in a portable mini-PC format for inspection, picking, and packing stations. This approach minimizes reliance on expensive GPUs or cloud services and ensures stable operation and enhanced security, even with unstable network conditions.

Park Chul-soo, CEO of Ourbox, emphasized the significance of this achievement, stating, "This accomplishment lays the groundwork for logistics companies to secure their own AI technology capabilities and solve on-site problems." He added, "By combining accumulated field experience with AI, we will intelligently transform overall logistics operations and implement practical AX that supports worker judgment."

“

By combining accumulated field experience with AI, we will intelligently transform overall logistics operations and implement practical AX that supports worker judgment.