KPMG Japan Launches AI Implementation Support Service for the Semiconductor Industry Using Proprietary Data

KPMG Japan (Chiyoda-ku, Tokyo; Co-Chairmen: Hiroyuki Yamada, Masahiko Chino) has begun offering a specialized AI implementation service for the semiconductor industry. This service leverages proprietary data, such as technical documentation, and tunes generative AI to suit corporate operational environments, aiming to build sustainable and highly productive business systems.

In the semiconductor industry, much of the information related to design, production technology, and quality management is stored as proprietary internal data. However, the retirement of veteran workers and the decline in the productive labor population make the transfer of skills and know-how a major challenge. While generative AI is effective for solving these issues, many companies have been reluctant to adopt it due to security concerns, as typical generative AI services rely on external data processing.

KPMG Japan, taking into account the business characteristics of the semiconductor industry, promotes the tuning and application of generative AI under a secure architecture that operates entirely within the corporate environment. This allows for highly professional and sensitive data, such as design drawings and technical documents, to be processed safely. This enables stable operations with a smaller workforce while transferring skills and know-how to the next generation.

This service provides one-stop support, from the formulation of use cases for productivity improvement and labor saving using proprietary data to defining expectations, gathering requirements, selecting and organizing corporate data (knowledge organization and data structuring), building a secure environment, AI tuning, and production deployment. This realizes a high-ROI AI environment with strong feasibility. Expected use cases include: "supporting skill transfer through rapid extraction and organization of technical information during troubleshooting/maintenance," "proposing alternatives or next steps for obsolete/scarce components," and "improving specification drafting efficiency and generating high-quality code."

[Service Features]

- Secure Architecture that Keeps Proprietary Data In-House: The generative AI is tuned within the client environment (on-premises or closed cloud, etc.), minimizing the risk of information leakage by keeping operations within a secure environment.

- High-Accuracy AI Optimized for Semiconductor Operations: The AI understands unique contexts, such as technical documents, work instructions, and design materials, aiming to improve response quality to levels usable in actual work environments.

- AI-Assisted Environment for Small-Team Operations: Simultaneously realizes standardization of tasks, reduction of workloads, and efficiency in skill transfer.

- Shorter Time and Lower Effort than Building from Scratch: Efficiently builds enterprise-specific AI by tuning based on existing models.