German agri-tech company Karevo has developed an AI-powered optical sorting system for potatoes and onions, targeting farms and packing operations seeking to reduce manual quality inspection.

The company, a spin-off from the Technical University of Munich (TUM), was founded in 2024 by Benedikt Keßler, Johannes von Wittke and Felix Beck. The technology originated from Keßler’s master’s thesis in mechanical engineering at TUM, which focused on automated quality control of unwashed table potatoes using computer vision and machine learning.

Karevo’s sorting systems use industrial cameras and AI-based image processing to identify defects and foreign material as potatoes move through the machine. The company lists greening, rot, mechanical damage, wireworm holes, growth cracks and scab among the defects that can be detected, alongside foreign bodies such as stones and soil clods.

According to Karevo, its systems achieve detection accuracy of approximately 95%. The company currently offers the Duo85, with a sorting capacity of up to five tonnes per hour, and the larger Trio110, capable of handling up to 10 tonnes per hour.

Each potato is photographed multiple times as it passes through the vision system, with the images processed in real time before pneumatic ejectors remove produce identified as defective.

The equipment has been designed so that it can be retrofitted into existing sorting lines, a feature intended to make optical sorting more accessible to smaller and medium-sized potato businesses that continue to rely heavily on manual inspection.

Karevo says the technology is suitable for table potatoes, seed potatoes and onions. The machines are manufactured in partnership with German potato equipment specialist EURO-JABELMANN Maschinenbau.

Karevo began developing its first prototypes through the UnternehmerTUM entrepreneurship ecosystem and entered the commercial market in 2025.