Kuken Industries, a major Japanese cooling tower manufacturer, has implemented an AI-powered search system to modernize access to its 70-year archive of technical drawings. Previously, locating specific installation plans required experienced guidance and could take half a day, especially for newer engineers unfamiliar with the archive's structure. The system, built in partnership with Recursive and using OpenAI models, allows engineers to search via part numbers, model names, or by uploading screenshots and PDFs.
This capability directly addresses the challenge of matching replacement parts to older towers where current models may not fit original installations. Since deployment, Kuken reports the tool can cut search time by approximately 90% annually. The technology not only accelerates part identification but also frees experienced engineers to mentor younger staff and focus on design improvements.
Newer engineers are now able to handle a wider range of jobs independently, reducing reliance on veteran guidance. Furthermore, Kuken is extending the AI tool to review drawings from partners, flagging specifications that deviate from requirements. This initiative represents a significant step in applying generative AI to preserve and utilize deep industrial knowledge, transforming a previously laborious archival process into a streamlined digital workflow that enhances both efficiency and innovation potential across the organization.
Source: OpenAI Blog · Summarized by HeadlinesBriefing