KIOXIA's innovative AiSAQ technology and SSDs bring AI-powered advances to logistics

KIOXIA's open-source AiSAQ is a game-changer as it offers an all-in-storage solution for AI by offloading vectors in large datasets from DRAM to SSD storage. Coupled with KIOXIA's Memory-Centric AI technology that stores AI training data on external storage, you've got AI-driven image recognition technology that could transform logistics.

2

In partnership with Tsubakimoto Chain Co. (Tsubakimoto Chain) and EAGLYS Inc. (EAGLYS), KIOXIA's AI image recognition technology can automatically identify products moving through complex logistics workflows. Designed for the growing e-commerce market, which sees large logistics networks handling higher volumes and a wider range of products, this scalable technology enables organizations to adapt to changing conditions while focusing on efficiency, cost management, and quality.

How it differs from traditional AI image recognition is simple. Those systems require tuning or even retraining for new products, including seasonal items, whereas KIOXIA AiSAQ and Memory-Centric AI store all new product data in high-capacity storage for quick retrieval, without the need to retrain the base model.

As mentioned in the introduction, this has the added benefit of mitigating the need to increase memory requirements as indexed data is moved to SSD storage. And that's where Retrieval Augmented Generation (RAG) enables fast and efficient retrieval. KIOXIA, Tsubakimoto Chain, and EAGLYS will showcase this new technology at the 2025 International Robot Exhibition, running this week from December 3 - 6 at Tokyo Big Sight.

The live demonstration will include products moving along a conveyor as the system captures image data and rapidly classifies them. KIOXIA AiSAQ open-source software is also available to download via GitHub.