News
Kuken Industries optimizes its drawing retrieval time from almost an hour to just 4.5 minutes by utilizing the OpenAI API, achieving a 90% increase in efficiency.
Kuken Industries, a prominent player in Japan’s cooling tower industry, has adopted an AI-enhanced search system to optimize its processes. By implementing the OpenAI API, the company has successfully decreased the time taken to find technical drawings for replacement components from nearly an hour to a mere 4.5 minutes. This advancement is expected to decrease the annual man-hours spent on searches by 90%, greatly improving operational productivity.
Challenges with Accessing Decades of Drawings
With a collection of over 70 years of cooling tower schematics, Kuken Industries encountered difficulties in retrieving specific documents, especially for newer engineers. Traditionally, finding the right drawing often necessitated consulting with seasoned colleagues or manually searching through digitized records. This task could range from several hours to half a day, causing delays in repairs and boosting operational expenses.
Development of the AI-Powered Search System
In partnership with Recursive, a Japanese AI startup and official associate of OpenAI, Kuken has designed a search system integrated with a field application. This system employs AI models to extract and organize data from the vast archive of drawings. Engineers can now search by part numbers, model names, or even screenshots, allowing for prompt access to pertinent documents. OpenAI’s models were chosen for their accuracy and cost efficiency, with the flexibility of the API facilitating easy integration into the application.
Strategic and Operational Benefits
The newly implemented system has revolutionized workflows at Kuken Industries. Engineers, regardless of their experience level, can now independently locate essential drawings, decreasing their reliance on senior colleagues. This speeds up repair times and enables junior engineers to focus on acquiring knowledge and enhancing problem-solving skills. Moreover, the system’s capability to match replacement parts with original installations cuts down the need for custom manufacturing, thus conserving time and resources.
Plans for Future Implementation
The AI-driven system is currently being piloted at Kuken’s Fukuoka facility, where it has already shown considerable time savings. Using data from 2025 operations, the company anticipates a 90% reduction in annual man-hours spent on drawing searches. Kuken aims to further enhance the system by improving its ability to interpret older, handwritten notes and ensuring substantial data security. The full-scale rollout across its Fukuoka and Chiba facilities is planned for late 2026.