Kuken Industries Deploys OpenAI API to Query 70-Year Drawing Archive in Minutes
Enterprise AI · MEP · Smart operations
Kuken Industries and AI startup Recursive developed an OpenAI API-based field app enabling engineers to locate replacement cooling tower drawings in 4.5 minutes, projecting a 90% annual reduction in drawing search time.
Japan-based cooling tower manufacturer Kuken Industries partnered with AI startup Recursive to implement an OpenAI API-powered search application across its 70-year archive of engineering drawings. The specialized equipment manufacturer, which supplies cooling towers to large facilities, shopping centers, and data centers, frequently customizes installations, complicating replacement part identification for newer personnel. The newly deployed field app allows engineers to search using drawing numbers, part identifiers, or uploaded screenshots and PDFs of parts lists. According to company metrics, the tool reduced the time required to retrieve specific replacement part drawings from nearly an hour down to 4.5 minutes, projecting an approximate 90% reduction in annual drawing search time. Kuken is also prototyping a drawing verification tool to automatically compare partner submissions against engineering specifications, reporting that 90% of prototype flags reflected genuine discrepancies.
Key takeaways
- Kuken Industries built an OpenAI API-based app with Recursive to search across a 70-year cooling tower drawing archive.
- Engineers can query via drawing numbers, part numbers, or uploaded images and PDFs of parts lists.
- Average drawing search duration dropped from nearly an hour to 4.5 minutes, with a projected 90% yearly time savings.
- Kuken is prototyping an AI review tool to identify discrepancies between partner drawings and specifications.
Source：OpenAI News Read the original ↗
Prepared, checked and automatically published by AI; not individually reviewed by a human.