Kuken Industries searches 70 years of drawings in minutes

Kuken Industries manufactures air conditioning and ventilation equipment, including cooling towers, and holds a major share of Japan's cooling tower market. When replacement parts are needed, engineers must consult a 70-year archive of drawings, but finding the right one was slow—a paper search could take half a day, and even the digitized archive required knowing which name or number to look up. Kuken partnered with Recursive, a Japanese AI startup and OpenAI partner, to build a search system and field app using OpenAI models, which performed best at extracting drawing details at reasonable cost. Engineers can now search by part number or by uploading an image or PDF of a drawing or parts list, and the app returns possible matches. Kuken expects this to cut drawing-search time by roughly 90% over a year, let newer engineers work more independently, free experienced staff for mentoring, and reduce printing costs. Kuken is also prototyping an AI tool to check partner drawings for discrepancies; about 90% of flagged issues in testing were genuine. The company aims to preserve engineers' accumulated decisions and pass that expertise to the next generation.

(Source：OpenAI)