Hitachi Launches AI Agent to Dramatically Streamline Quality Assurance Operations as Part of HMAX Industry

Hitachi, Ltd. (hereinafter, Hitachi) has started offering an AI agent (hereinafter, this AI agent) called the 'Quality Knowledge System' as part of its HMAX Industry lineup from April 2026. This AI agent derives optimal insights from vast amounts of past quality-related data, significantly streamlining quality assurance operations in the manufacturing industry. This AI agent supports the series of tasks undertaken by the quality assurance department when troubles such as equipment failures occur, including searching past trouble response records and creating response reports. This significantly reduces the time required for information gathering in quality assurance operations, enabling faster responses to customer inquiries and improving service quality. In the future, Hitachi plans to add functions for analyzing quality status, supporting the formulation of defect prevention measures, and contributing to the improvement of product quality itself. Furthermore, the company is considering collaboration with the OT (Operational Technology) domain in the future, aiming for deployment as a Physical AI.

Specifically, based on a database of accumulated past cases, Hitachi analyzes business processes such as what keywords and sequences experts used to search for information and make judgments. By formalizing the tacit knowledge accumulated as know-how, the company digitally reproduces judgments that previously depended on the experience and intuition of veteran engineers. This enables quick and accurate decision-making regardless of the experience level of the person in charge. As a digital service combining Hitachi's long-accumulated domain knowledge with advanced AI, it contributes to solving manufacturing industry challenges such as skill transfer and productivity improvement.

This AI agent was first introduced at Hitachi's Omika Works (Hitachi City, Ibaraki Prefecture), a development base for social infrastructure control systems, as part of a 'Customer Zero' initiative to verify its effectiveness by using it internally first. As a result, the company confirmed effects such as a reduction in trouble response case search time by approximately 90%, report creation time by over 80%, and defect cause analysis time by over 80%. Based on these results, Hitachi will implement application designs tailored to customers' quality assurance business processes, supporting the smooth introduction and establishment of this AI agent.

Going forward, Hitachi will actively deploy this AI agent to the manufacturing industry as part of its HMAX Industry lineup. By utilizing the knowledge model (knowledge) gained from this AI agent, including engineer insights, defect information, and customer inquiry data, the company aims to establish a virtuous cycle of feeding back on-site wisdom and inspiration to the upstream processes of manufacturing. In the future, Hitachi plans to deploy similar services not only for quality assurance operations but also for design, manufacturing, and maintenance operations.

Hitachi is focusing on 'HMAX Industry,' a next-generation solution suite that embodies Lumada 3.0, combining product-rich installed base (digitalized asset) data with domain knowledge and advanced AI for the industrial sector. Aiming to become a leading company in Physical AI, Hitachi contributes to realizing a prosperous society by maximizing customer lifetime value and transforming industries globally through the provision of 'Industrial Solutions' centered on these technologies.

Main Features of this AI Agent

The core technology of this AI agent applies the data management and AI utilization know-how of Hitachi's specialized organization, the 'Generative AI Center.' By incorporating over 100 pieces of expert business knowledge into the AI and repeatedly improving and tuning prompts, Hitachi has built a unique AI agent equipped with an expert's perspective. Specifically, it achieves the following three highly effective business supports:

1. High-precision search that allows even inexperienced staff to draw on expert know-how

It can search past trouble response records and manuals with high precision based on natural language questions or the text of customer inquiry emails. By formalizing the judgment process for events that previously depended on expert experience and intuition (tacit knowledge) and incorporating it into the AI agent, even inexperienced staff can respond quickly, reducing search time by approximately 90%.

2. Improved customer satisfaction through rapid creation of high-quality draft reports

By simply inputting keywords such as event, phenomenon, and countermeasure policy, even inexperienced staff can quickly create professional and easy-to-read draft trouble response reports. This eliminates the personalization and quality variation in report creation, enabling faster reporting and provision to customers, leading to improved customer satisfaction and reducing report creation time by over 80%.

3. Improved trouble response capability and product quality through multifaceted quality status analysis (function to be added in the future)

This will enable the AI agent to perform multifaceted analysis not only on accumulated vast quality documents such as trouble response reports but also on information like functional specifications. This efficient analysis, independent of the individual perspective of the person in charge, will not only support the formulation of defect prevention measures but also contribute to improving product quality itself by feeding back analysis results to the development department, reducing analysis time by over 80%.

Background

In the environment surrounding the manufacturing industry, in addition to global labor shortages and generational shifts, products and systems are becoming more sophisticated and complex, making the knowledge and skills required of companies increasingly specialized and diverse. Particularly when unexpected situations such as equipment failures or quality troubles occur, quick and accurate responses by the quality assurance department are becoming more important from the perspectives of business continuity and customer satisfaction. On the other hand, in actual practice, many situations in trouble response and cause analysis depend on the experience and judgment that expert engineers have cultivated over many years. This creates challenges in passing on skills and know-how and achieving efficient and stable quality assurance operations that are not affected by the experience level of the person in charge.