Technology Record - Issue 32: Spring 2024

111 INDUSTRIALS & MANUFACTURING CASE STUDY: DENSO Denso Corporation, Japan’s largest manufacturer of automotive components, has several initiatives in progress relating to digitalisation and digital experience (DX) as it seeks to make full use of its human strengths and technical knowledge. Its Processing DX Laboratory accelerates research and development of new processing solutions using digital technology. A primary challenge for the lab relates to data. General manager Junya Asaoka explains: “The information and data required are scattered around and we had to rely on experience and manual work when collecting the data based on its location. In addition, to utilise the data, we had to manually structure it in a variety of media and formats, including paper and digital.” Denso turned to Avanade for support, and a workshop was organised. Koichi Tanaka, a manager in the Denso team, says: “We examined the creation of an environment for collecting, analysing and accumulating knowledge to identify actual on-site problems and issues. We found a path to solving these issues.” Avanade posed the question, ‘Can we verify the potential of AI?’, and developed a proof of concept (PoC) to automate tasks that were previously carried out manually. The PoC uses AI to verify the structuring of unstructured data, such as technical information from the field. Information in various formats – such as handwritten, PDF and Microsoft Excel files – needed to be standardised; some was structured data, while some was unstructured data such as natural language. The next step was to verify whether AI-based structuring could achieve the required quality, allowing users to search for the information they want, when they want it, by adding tags to the data and cataloguing it. The resulting system combines Avanade Insight Discovery and Microsoft Azure OpenAI. “We were concerned about the difficulty in meeting internal security standards with the services provided by OpenAI and the secondary use of input data through the nonAPI ChatGPT,” says Daichi Murakami, from the Denso team. “However, Avanade explained that data is only stored in Avanade Insight Discovery with the key items required for the search extracted. GPT is used as an analysis resource, so the system ensures security. “Through this PoC, we understand that it is possible to promote searches and information utilisation. The aim is to create a system like an ‘in-house ChatGPT’, and the PoC has allowed us to verify the feasibility of such a system.” Asaoka adds: “AI use will be even more important in the future. We are looking forward to having Avanade as a partner to plan and study AI use while constantly sharing new information on the latest technologies and initiatives.” Japanese auto parts manufacturer Denso has completed a proof of concept with Avanade to verify the potential of an ‘in-house ChatGPT’ Making sense of data with AI Avanade helped Denso verify that AI-based structuring of data could achieve the required quality for effective search Photo: Denso