Kuken Industries is using the OpenAI API to search about 70 years of cooling tower drawings, and the company says engineers can now find the drawing needed for a repair in roughly 4.5 minutes, down from nearly an hour. The archive covers decades of engineering documentation, and locating the correct drawing quickly is a routine bottleneck when equipment needs to be fixed. The deployment is presented as a customer story on OpenAI's website, which positions it as an example of large language model tools being applied to legacy technical records. The show notes do not give details such as the models used, the number of documents indexed, implementation costs, or how accuracy was measured, so the figures should be read as the company's own account of the results. For Australian businesses, the practical takeaway is limited. Organisations that run industrial plant, cooling towers or other facilities with large engineering archives may find the approach worth a small pilot, with costs driven by document volume and API usage. Where drawings or related records contain personal information, such as contractor names or signatures, the Privacy Act 1988 and the Australian Privacy Principles would apply to how that data is processed. For most other businesses, this case changes nothing immediately.