Rolftech, with the help of language models and service experts, organized Rolf's directory, which had been forming for 35 years. During this time, it accumulated almost 475 thousand entries – with abbreviations, different translations, and several names for the same job.
The problem was that an experienced mechanic could still understand an entry like "NAR HANDLE DOOR SWING R FR S/U", but for a new employee or a digital system, such formulations are practically useless. After removing duplicates, 247.6 thousand unique operations remained, and for 238 thousand of them, experts have already confirmed the correct classification.
AI suggested which car component each entry referred to, and service advisors and engineers checked the results. Their feedback was returned to the system, so each subsequent pass became more accurate.
As a result, instead of a set of internal abbreviations, a single logic is formed: for example, "Car → Engine → Lubrication System → Oil Filter", and the work itself receives a clear name like "Oil Filter + Replacement". Suitable spare parts and a specific car configuration can be immediately linked to it.
For the service, this means faster search and less dependence on whether the employee knows the old internal designations. For the client, it provides a basis for a clearer repair record, accurate selection of work and parts, and a unified service history. The new directory is planned to be implemented into the Flora platform in December 2026.