EMNI and Toyo Kohan Launch Joint Project to Overcome Knowledge Dependency in Maintenance Operations with Generative AI 'AI Interviewer'
EMNI Inc., an AI startup from Kyoto University's Matsuo Lab, has partnered with Toyo Kohan Co., Ltd. to launch a project to advance maintenance operations using a generative AI-powered 'AI Interviewer.' The initiative aims to overcome the reliance on the experience and intuition of veteran staff in manufacturing. The AI will conversationally extract tacit knowledge from experts and systemize scattered past reports, enabling anyone to respond quickly and effectively, thereby minimizing production downtime.
📋 Article Processing Timeline
- 📰 Published: May 19, 2026 at 19:00
- 🔍 Collected: May 19, 2026 at 10:31
- 🤖 AI Analyzed: May 20, 2026 at 06:50 (20h 18m after Collected)
1. Overview
EMNI Inc. (Headquarters: Chiyoda-ku, Tokyo; CEO: Yuta Shimono; hereinafter 'EMNI'), in collaboration with the DX Promotion Department of Toyo Kohan Co., Ltd. (Headquarters: Shinagawa-ku, Tokyo; hereinafter 'Toyo Kohan'), has initiated a joint project to formalize know-how in the maintenance department and minimize manufacturing downtime by utilizing a generative AI-powered "AI Interviewer." This project aims to achieve 'de-personalization' (overcoming dependency on specific individuals) in maintenance operations by merging EMNI's generative AI technology with Toyo Kohan's extensive manufacturing and maintenance knowledge.
2. Background and Challenges
Maintenance work, which supports the stable operation of manufacturing sites, requires prompt responses to sudden equipment failures. However, the quality of this response is often heavily dependent on the experience and intuition of veteran maintenance personnel. At Toyo Kohan, the following challenges had become apparent:
● Tacit Knowledge of Veterans: While data such as maintenance histories were being stored, the background information—such as 'why a particular response was chosen' or 'what was the thought process behind a decision'—was not recorded. This left critical decision-making processes confined to the minds of veterans.
● Bottlenecks from Veteran Dependency: The resources of a small number of veteran maintenance staff were consumed with each incident, limiting the overall response speed of the organization.
● Scattered Past Cases: Trouble reports and work procedure manuals were in various formats and stored in different locations, making it impossible to immediately reference past knowledge when similar troubles recurred.
3. Project Purpose and Outline
This project aims to use generative AI to overcome knowledge dependency in maintenance operations. Specifically, it will structure both the tacit knowledge in veterans' minds and the explicit knowledge scattered throughout the organization into a format that AI can handle. This will create a state where everyone can access standardized, high-quality knowledge. The ultimate goal is to enable anyone to respond optimally and swiftly in maintenance situations.
● Verbalizing Tacit Knowledge: The AI Interviewer will conduct in-depth interviews with veteran maintenance staff. Through dialogue, it will systematically extract insights, decision criteria, and rules of thumb that are difficult for humans to comprehensively document, converting them into explicit knowledge that anyone can use.
● Organizing and Standardizing Explicit Knowledge: Generative AI will systematize scattered past trouble reports, work procedure manuals, and maintenance histories. This will enable anyone to make appropriate response decisions based on past knowledge, even for sudden equipment failures.
Specific Initiatives:
1. Organizing Past Data (Structuring Explicit Knowledge): Using EMNI's proprietary technology, which includes a data structuring platform capable of handling not only Word and PDF files but also electrical circuit diagrams, we will organize and structure trouble reports and procedure manuals from various formats and locations into an AI-friendly format. Leveraging EMNI's expertise specialized in the manufacturing industry, we will tailor the output to match Toyo Kohan's workflow and terminology for practical on-site use.
2. Formalizing Tacit Knowledge (Introducing the AI Interviewer): The AI will conduct in-depth interviews with veterans, systematically organizing insights, decision criteria, and rules of thumb that are difficult for humans to comprehensively document.
3. Building a Knowledge Utilization Platform: The AI will appropriately search and present the formalized tacit knowledge and the organized explicit knowledge. The platform will be customized to Toyo Kohan's operational flow so that necessary past knowledge can be accessed immediately when needed on the maintenance floor.
4. Expected Effects
The project is expected to yield the following benefits:
● Standardization and Acceleration of Incident Response: Enables anyone to quickly judge and execute the appropriate response to sudden failures, reducing line recovery time.
● Minimization of Manufacturing Downtime: Rapid access to necessary knowledge and standardized responses will minimize line stoppages and improve productivity.
● Resolution of Knowledge Dependency: Moves away from reliance on veterans, boosting the overall response capability of the maintenance department. This accelerates skill transfer and helps younger employees become effective more quickly.
● Digital Assetization of Knowledge: On-site knowledge that was previously tacit becomes a valuable 'company asset' that can be utilized across the entire organization.
5. Future Outlook
Moving forward, we will verify the system's effectiveness through a Proof of Concept (PoC) and proceed with a full-scale rollout for practical operation. In the future, we plan to promote the AI transformation of Toyo Kohan's entire manufacturing process, considering expansion to other areas beyond maintenance, such as manufacturing, quality, and design.
About Toyo Kohan Co., Ltd.
Since its establishment in 1934, Toyo Kohan Co., Ltd. has been manufacturing and selling products such as tinplate, thin sheet, various surface-treated steel sheets, and functional materials, leveraging its proprietary technologies in rolling, surface treatment, and lamination.
EMNI Inc. (Headquarters: Chiyoda-ku, Tokyo; CEO: Yuta Shimono; hereinafter 'EMNI'), in collaboration with the DX Promotion Department of Toyo Kohan Co., Ltd. (Headquarters: Shinagawa-ku, Tokyo; hereinafter 'Toyo Kohan'), has initiated a joint project to formalize know-how in the maintenance department and minimize manufacturing downtime by utilizing a generative AI-powered "AI Interviewer." This project aims to achieve 'de-personalization' (overcoming dependency on specific individuals) in maintenance operations by merging EMNI's generative AI technology with Toyo Kohan's extensive manufacturing and maintenance knowledge.
2. Background and Challenges
Maintenance work, which supports the stable operation of manufacturing sites, requires prompt responses to sudden equipment failures. However, the quality of this response is often heavily dependent on the experience and intuition of veteran maintenance personnel. At Toyo Kohan, the following challenges had become apparent:
● Tacit Knowledge of Veterans: While data such as maintenance histories were being stored, the background information—such as 'why a particular response was chosen' or 'what was the thought process behind a decision'—was not recorded. This left critical decision-making processes confined to the minds of veterans.
● Bottlenecks from Veteran Dependency: The resources of a small number of veteran maintenance staff were consumed with each incident, limiting the overall response speed of the organization.
● Scattered Past Cases: Trouble reports and work procedure manuals were in various formats and stored in different locations, making it impossible to immediately reference past knowledge when similar troubles recurred.
3. Project Purpose and Outline
This project aims to use generative AI to overcome knowledge dependency in maintenance operations. Specifically, it will structure both the tacit knowledge in veterans' minds and the explicit knowledge scattered throughout the organization into a format that AI can handle. This will create a state where everyone can access standardized, high-quality knowledge. The ultimate goal is to enable anyone to respond optimally and swiftly in maintenance situations.
● Verbalizing Tacit Knowledge: The AI Interviewer will conduct in-depth interviews with veteran maintenance staff. Through dialogue, it will systematically extract insights, decision criteria, and rules of thumb that are difficult for humans to comprehensively document, converting them into explicit knowledge that anyone can use.
● Organizing and Standardizing Explicit Knowledge: Generative AI will systematize scattered past trouble reports, work procedure manuals, and maintenance histories. This will enable anyone to make appropriate response decisions based on past knowledge, even for sudden equipment failures.
Specific Initiatives:
1. Organizing Past Data (Structuring Explicit Knowledge): Using EMNI's proprietary technology, which includes a data structuring platform capable of handling not only Word and PDF files but also electrical circuit diagrams, we will organize and structure trouble reports and procedure manuals from various formats and locations into an AI-friendly format. Leveraging EMNI's expertise specialized in the manufacturing industry, we will tailor the output to match Toyo Kohan's workflow and terminology for practical on-site use.
2. Formalizing Tacit Knowledge (Introducing the AI Interviewer): The AI will conduct in-depth interviews with veterans, systematically organizing insights, decision criteria, and rules of thumb that are difficult for humans to comprehensively document.
3. Building a Knowledge Utilization Platform: The AI will appropriately search and present the formalized tacit knowledge and the organized explicit knowledge. The platform will be customized to Toyo Kohan's operational flow so that necessary past knowledge can be accessed immediately when needed on the maintenance floor.
4. Expected Effects
The project is expected to yield the following benefits:
● Standardization and Acceleration of Incident Response: Enables anyone to quickly judge and execute the appropriate response to sudden failures, reducing line recovery time.
● Minimization of Manufacturing Downtime: Rapid access to necessary knowledge and standardized responses will minimize line stoppages and improve productivity.
● Resolution of Knowledge Dependency: Moves away from reliance on veterans, boosting the overall response capability of the maintenance department. This accelerates skill transfer and helps younger employees become effective more quickly.
● Digital Assetization of Knowledge: On-site knowledge that was previously tacit becomes a valuable 'company asset' that can be utilized across the entire organization.
5. Future Outlook
Moving forward, we will verify the system's effectiveness through a Proof of Concept (PoC) and proceed with a full-scale rollout for practical operation. In the future, we plan to promote the AI transformation of Toyo Kohan's entire manufacturing process, considering expansion to other areas beyond maintenance, such as manufacturing, quality, and design.
About Toyo Kohan Co., Ltd.
Since its establishment in 1934, Toyo Kohan Co., Ltd. has been manufacturing and selling products such as tinplate, thin sheet, various surface-treated steel sheets, and functional materials, leveraging its proprietary technologies in rolling, surface treatment, and lamination.