AI-Powered Product Information Management – Fujitsu Limited today announced that it has implemented a multi-AI agent infrastructure for the product information management business of MISUMI Group Inc. (MISUMI). Building on the operation platform of its business model Uvance [1], Fujitsu leveraged “Fujitsu Data Intelligence PaaS” [2] to establish a system that supports the entire process, from data collection and pre-processing to AI-driven generation, evaluation, and final human verification. MISUMI began operating this new workflow in October.
This workflow targets tens of thousands of products listed on MISUMI’s e-commerce site. By creating an operational model where humans and AI collaborate, the project aims to improve both the efficiency and quality of product information management.
The platform was built in approximately 2.5 months. This rapid deployment was achieved through an agile development process, where Fujitsu’s Forward Deployed Engineers (FDE) worked closely with MISUMI’s business personnel to integrate professional operational judgment criteria and verification methods directly into the AI agents.
Background
As MISUMI has expanded its product offerings, the volume of data requiring registration, including product category classifications and technical specifications, has grown substantially. Registering new products and series typically required manual, time-consuming processes across multiple internal and external systems. These manual dependencies created bottlenecks, delaying product launches on the e-commerce site and presenting challenges in maintaining data consistency and quality. To address these issues, Fujitsu is leveraging Fujitsu Data Intelligence PaaS to transform the business processes through AI integration.
Figure : Multi-AI agent utilization in product information management business
Project Highlights
1. Rapid implementation
Fujitsu quickly established a data utilization foundation and AI execution environment by leveraging Fujitsu Data Intelligence PaaS. In addition, Fujitsu’s Forward Deployed Engineers (FDE) worked in close cooperation with business personnel at MISUMI to incorporate their expertise, such as operational judgment criteria and verification methods, into the AI agents. The team also built a workflow that integrates a confirmation and approval process by human operators. Through this agile development approach, the multi-AI agent platform was successfully applied to actual business operations in just 2.5 months after the project began.
2. Quality enhancement through AI collaboration
The workflow utilizes nine AI agents, each with distinct roles such as generating series names, suggesting categories, extracting specifications, and checking for restricted characters. These agents mutually evaluate the validity of their outputs, while final human approval ensures accuracy and consistency across the catalog.
3. Focus on high-value operations
Automating routine registration and verification tasks significantly reduces the administrative burden on staff. This allows employees to shift their focus toward higher-value activities, such as strategic decision-making and business process improvements.
Future Plans
Fujitsu will continue to collaborate with MISUMI to gradually expand the scope of the developed data models and AI-driven workflows. The next phases will include end-to-end management from initial product data ingestion from suppliers to master data management, with the goal of further automating the entire workflow through integration with peripheral systems. Fujitsu remains committed to supporting business transformation in the manufacturing sector through its Uvance model, driving faster time-to-market and increased industrial competitiveness.
For more information visit: global.fujitsu.
