Hitachi, Ltd. (Hitachi) has introduced an AI-powered sales system designed to enhance the efficiency of commodity invoicing management. What makes this system unique is that it doesn’t require prior data input related to product transactions. Instead, the AI learns and improves on its own through self-study, enabling more efficient invoice management and accelerating the pace of commercial transactions. The system targets businesses looking to optimize their inventory control.
Compared to traditional manual order processing, Hitachi’s AI sales system can reduce losses from overstocking or stockouts by up to 75%. To demonstrate the effectiveness of the system, Hitachi used a beer production simulation across four stages: brewing, wholesale, retail, and distribution. By applying AI to stabilize supply and demand for beer, the system helps minimize operational losses.
Unlike conventional AI systems that rely on large pre-labeled datasets, Hitachi's AI leverages data generated by AI agents running in the system. These agents compete against each other, continuously learning and improving their performance. Each agent goes through a cycle of trial, observation, and learning, gradually refining its approach to invoicing and reducing business losses. This competitive environment fosters innovation and better decision-making.
If an AI agent produces a model that leads to significant losses, it will adjust and refine its strategy until the issue is resolved. Additionally, the system evaluates the performance of all agents, discarding underperforming ones and creating new, improved versions. This self-improving mechanism allows the AI to optimize the entire supply chain without requiring additional data inputs.
This approach, where AI competes and evolves independently, is similar to the method used by AlphaGo Zero, which outperformed earlier versions like AlphaGo. By constantly challenging itself, the AI enhances its learning capabilities, leading to smarter decisions and greater efficiency—just like how AlphaGo Zero became the top Go champion in the world.
Langrui Energy (Shenzhen) Co.,Ltd , https://www.langruibattery.com