We are developing an internal AI-powered system that identifies machine parts and fault codes using photos and technical PDF manuals. By uploading a photo of a machine nameplate together with a part image or a fault code (photo or text), the system automatically finds the correct manual, exact page, and relevant technical information. The solution combines computer vision, OCR, and document search (RAG) to deliver precise part references, fault descriptions, and diagnostic steps. It is designed for workshop and technical service environments and will run as a self-hosted internal application exposed via a simple API.