CannAI Solutions UG (haftungsbeschränkt)

CannAI Solutions UG (haftungsbeschränkt) is a software company in Eckernförde, Schleswig-Holstein, building computer vision for plant diagnostics.

A grower photographs a plant. Our models separate healthy tissue from disease, nutrient deficiency and pest damage, check the finding against cultivation data and an agronomic rule base, and return a recommendation together with a confidence value. Where that confidence is not sufficient, the system asks for a new photograph instead of returning a class. We also measure the direction of our errors, not only how often we are right. Advice that points the wrong way, for example more fertiliser where the actual cause is a pathogen, is the expensive failure in the field, and accuracy alone hides it.

Inference runs on our own GPU cluster in our own company rooms in Germany, so grower imagery stays inside infrastructure we own and operate. From October 2026 we additionally use a GPU-hour allocation at the AI Service Centre WestAI (RWTH Aachen), granted after technical review, for three transfer studies: disease recognition under laboratory to field domain shift, weed recognition across fields, seasons and camera platforms, and open-set insect recognition that separates pests, beneficials and pollinators. Data splits are frozen and the evaluation code will be published.

Our product has been in the Apple and Google app stores since June 2026 with paying customers. Our first market is legal cannabis cultivation in Germany, where we are an ancillary software and data provider and do not grow, handle or sell plants. That market gives us a continuous stream of labelled real-world images. The product is at TRL 7 in this first market, the transfer to horticulture and arable crops is at TRL 5. The stack is crop agnostic: extending it to a further crop or to a partner product needs labelled images and a training run, not a new system.

What we are looking for at Smart Agri Connect: greenhouse and field sites willing to validate the system on their own crops, technology providers with hardware or an agronomy platform but no in-house vision team, and partners for Horizon Europe Cluster 6 and Eurostars, preferably in the Netherlands, Denmark or Belgium.