What challenges did Taranis face before AI implementation?

Help farmers monitor huge farmland areas and reduce crop yield loss caused by insects, disease, weeds, and nutrient deficiencies. Handle high-volume drone image uploads from remote locations. Train and serve machine learning models at scale.

What was the impact of the AI solution?

Reduced upload times from up to a day to several hours. Scaled GPU capacity for training. Reduced cost per photo taken by about 10x. Allowed features to be released almost continuously. Enabled earlier intervention to prevent crop loss.

Source

GCPExpanded

Taranis Case Study | Google Cloud

Use case typePredictive decision supportUpdated Jun 13, 2026

Taranis is a precision agriculture intelligence platform that helps farmers monitor their fields and reduce crop yield loss. The company uses Google Cloud to enable high-volume drone image uploads and train TensorFlow machine learning models for detecting disease, pest, and abiotic stress from agricultural imagery. Using Google Cloud, Taranis supports large-scale image processing and serves insights through dashboards so farmers can intervene earlier and target solutions.

Organization: Taranis
Industry: Agriculture
Location: Israel
Published: January 2024

Reported outcomes

10x

cost_per_photo_reductionCost savings

4,000 V100 GPUsgpu_scale_out

Strategic outcomes

Speed & agilityEnabled near-continuous feature releasesCost efficiencyLowered cost per photoScale & capacityScaled machine learning training capacityCustomer experience & trustEnabled earlier crop intervention

Catalog median for cost savings deployments: −45% across 345 reported metrics. Compare benchmarks →

Primary read

Use case focus

Showing 3 of 3

1Computer Vision
2Predictive Analytics
3Data Pipeline

Help farmers monitor huge farmland areas and reduce crop yield loss caused by insects, disease, weeds, and nutrient deficiencies.
Handle high-volume drone image uploads from remote locations.
Train and serve machine learning models at scale.

Migrated to Google Cloud for scalable storage, processing, and model training.
Used Compute Engine, Kubernetes Engine, Cloud SQL, and Cloud Pub/Sub to support a scalable drone- and satellite-image pipeline.
Trained TensorFlow models on millions of tagged image data points to detect disease, pest, and abiotic stress.
Delivered insights through dashboards to help farmers act earlier and with more targeted interventions.

Technologies

Compute Engine Kubernetes Engine Cloud SQL Cloud Pub/Sub TensorFlow

Reduced upload times from up to a day to several hours.
Scaled GPU capacity for training.
Reduced cost per photo taken by about 10x.
Allowed features to be released almost continuously.
Enabled earlier intervention to prevent crop loss.

Architecture

Taranis migrated its drone- and satellite-image processing workflow to Google Cloud. The architecture uses Compute Engine GPU capacity for image processing and TensorFlow model training, Kubernetes Engine for containerized workloads, Cloud SQL for database storage, and Cloud Pub/Sub to support the scalable pipeline that ingests large uploads from remote locations and serves derived insights to customers.

Implementation partners1

DoiT International

Sources & evidence1

ExpandedExpanded

The same organization appears in newer AI deployment evidence.