Business situation
Japan's largest global manufacturer of automotive components in the Toyota Group, DENSO Corporation (DENSO), is working to connect people and transform work by making full use of its strengths in human resources and workplace technology knowledge.
The information and data required for the research and development of process technologies is scattered around, requiring manual work to collect and structure it. To mitigate this issue, DENSO asked Avanade for support to help it create an environment for collecting, analyzing and accumulating knowledge to identify on-site problems.
Solution
We proposed a proof of concept (PoC) to address the challenges of data accumulation, collection and format conversion. The solution automates tasks that were previously carried out manually and then uses AI to verify the structuring of unstructured data, such as technical information from the field.
The PoC combines Microsoft Azure OpenAI and Avanade Insight Discovery, which is a data analysis platform that runs on Azure and uses natural language processing technology. Specifically:
- DENSO’s knowledge is stored in Avanade Insight Discovery, catalogued as data.
- Azure OpenAI’s GPT is used as an analysis resource and search tool to take advantage of the AI model’s wealth of knowledge and information.
“Through the PoC, we have validated several AI models, including GPT-3 and GPT-3.5, and have verified that Avanade can automatically structure models that previously had to be done manually.”
Results
Tasks that were previously carried out manually can now be automated, which is expected to reduce hours of work for employees. By combining Avanade Insight Discovery, it’s possible to accumulate the knowledge of the company and promote search and information utilization that better meets the company's needs.
Regarding future development, DENSO leaders are considering how to apply this solution to business operations. They’re currently verifying the demonstrated effect using sample data to see how much search accuracy changes between raw data and structured data to create an in-house chatbot.
Find out how a human-centered approach to AI can deliver results for your business.