Reparo

When an industrial motor fails, the production line stops and every hour of downtime costs money. The mechanic photographs the nameplate and sends it to a parts supplier, where someone manually reads the photo, searches the catalog, and — if the motor is discontinued — tries to find a compatible replacement by specification. This takes hours, and the urgent order goes to whoever answers first. Reparo turns this into seconds. Upload a photo of a motor nameplate and an AI vision model running on AMD GPUs (via AMD Token Factory) extracts the manufacturer, model, power, speed, frame size and mounting. Reparo searches the supplier catalog for the exact part, and when it is discontinued or out of stock, it finds a compatible replacement by specification — explaining exactly why it fits. The intelligence is in the matching: anyone can search by part number, but Reparo answers the question a database cannot — "this code no longer exists, what actually fits instead?" Crucially, it refuses unsafe substitutes: a motor with a different mounting type is never suggested, because in mechanics "close enough" means "doesn't fit". When nothing truly matches, Reparo says so honestly rather than risk a wrong part. Built for the Reinvent Commerce track: faster response means more won orders for the supplier and less downtime for the factory.