How AI is transforming Kazakhstan’s agriculture

AI generated technologies are being tested in Kazakhstan where to monitor crops, determine optimal harvesting times and identify signs of disease, El.kz reports.

In Kazakhstan, systems are being developed that can survey crops using satellite imagery, direct drones to areas of concern and assess plant maturity. Some have undergone initial field trials, while others are still being tested in laboratory conditions. At the same time, digital solutions for monitoring soil conditions and crops are already being used on farms.

Detecting a disease before visible symptoms appear

The first signs of infection often develop within plant tissues. Water content, chlorophyll concentration and photosynthetic activity begin to change. However, the leaves may retain their usual colour and appear completely healthy.

Hyperspectral cameras can help detect such changes. They capture light reflected by plants across dozens or even hundreds of narrow spectral bands. The resulting data are analysed by a neural network trained to distinguish the characteristic features of healthy and infected plant tissues.

In 2026, the journal Smart Agricultural Technology published a review of 79 studies on the early detection of plant diseases. Under controlled conditions, the average disease recognition accuracy reached 91.4%.

In open-field conditions, this figure fell to 86.3%. Sunlight, shadows, dust and variations in vegetation make analysis more challenging. Furthermore, some infections produce spectral changes similar to those caused by drought or nutrient deficiencies.

Researchers view this type of diagnostic technology as an opportunity to detect plant diseases before visible symptoms emerge. How much earlier an infection can be identified depends on the pathogen, crop type and equipment used.

A satellite identifies a suspicious area. A drone investigates it

Satbayev University is developing Agroscope, a system that integrates satellite imagery, unmanned aerial vehicles (UAVs) and artificial intelligence. The project is led by Ravil Mukhamediev.

The process begins with satellite-based monitoring of agricultural land. The software calculates the Normalized Difference Vegetation Index (NDVI), which is used to assess vegetation health based on reflected light. If a particular area differs significantly from the rest of the field, the system flags it for further inspection.

A drone is then dispatched to the selected location. Its cameras capture higher-resolution images, allowing the condition of the crops to be examined in greater detail. Potential causes of the detected anomalies include insufficient soil moisture, pest damage, plant diseases and uneven crop development.

The developers intend to use Agroscope for crop monitoring, fertiliser application planning and yield forecasting. Kazakhstan is also considering the localisation of AI-enabled agricultural drone manufacturing. The Agroscope system itself is still under development.

Diagnosing plant diseases from a photograph of a leaf

For certain applications, an ordinary smartphone may be sufficient. Once a disease has produced visible symptoms, computer vision technology can identify its characteristic features from a photograph.

In October 2026, students Danil Rush and Zhangir Satmyrza from Akhmet Baitursynuly Kostanay Regional University presented a technology for diagnosing diseases affecting agricultural crops. The project was showcased at the AI & Digital Bridge forum in Astana.

The software analyses an image of a plant and identifies potential signs of disease. The developers have also incorporated recommendations to help users determine what measures should be taken.

However, the visual appearance of a leaf does not always make it possible to establish the underlying cause of damage. Yellowing, for example, may result from infection, nutrient deficiencies or improper irrigation. Consequently, computer-generated assessments require verification by a specialist. No data on the accuracy of the Kostanay project in large-scale field trials are currently available.

Why it is still too early to retire agronomists

The main challenge facing digital diagnostic systems is the wide variety of conditions under which agricultural crops are grown. An algorithm that reliably identifies a disease in laboratory images may produce errors under different lighting conditions, when applied to a new crop variety or when several stress factors occur simultaneously.

The quality of training data is also critical. For a system to reliably identify diseases affecting wheat grown in Kazakhstan, it must be validated using local varieties, weather conditions typical of the region and different stages of infection development.

Cost is another consideration. Hyperspectral cameras, drones, sensors and image-processing systems require investment. For agricultural enterprises, the decisive factor will be the economic return: how much can be saved on crop treatments, irrigation and fertilisers, and what proportion of potential crop losses can be prevented?

Developments in Kazakhstan are already making it possible to address specific agricultural challenges. Near Karaganda, drones are being used to assess crop conditions; in Northern Kazakhstan, a neural network is being used to evaluate the maturity of forage crops; and students in Kostanay are testing computer-based plant diagnostics.

The next stage for these projects will be extended field trials to establish how accurately the algorithms perform under real farming conditions and whether their use is economically viable.