SMBSC Modernizes Beet Sugar Intake Quality Screening with AI on AWS

Customer

Southern Minnesota Beet Sugar Cooperative (SMBSC) is a grower-owned agricultural cooperative headquartered in Minnesota. The cooperative manages beet sugar intake across a broad grower network and must process a high volume of seasonal truckloads during a narrow harvest window.

The Challenge

SMBSC needed a more consistent way to inspect incoming beet loads for impurities such as dirt and leafy greens. Manual visual grading introduced several operating constraints:

• - Binary greens present/not present assessment limited precision
• - Manual review made it difficult to inspect every load consistently during peak harvest
• - Variable lighting and weather conditions complicated image-based inspection
• - Delayed visibility into quality trends increased financial and operational risk

The Solution

Tactical Edge AI worked with SMBSC to build a near-real-time computer vision scoring system on AWS. The system replaced binary manual inspection with continuous 0-100 quality scoring, helping operations teams identify impurity levels quickly and consistently.

The solution combined onsite process assessment, camera setup, data pipeline design, model training, model deployment, and alerting. Key AWS services included:

• - Amazon SageMaker AI for model training and deployment
• - Amazon S3 for intake image storage and event-driven processing
• - AWS Lambda for serverless backend orchestration and dashboard updates

When impurity levels exceeded configured thresholds, the system generated alerts for operations leaders, shortening response time and supporting more consistent quality decisions across the intake process.

Results

The AI-powered screening system delivered measurable improvements in accuracy, speed, and operational visibility:

• - More than 90 percent impurity detection accuracy
• - Load grading in 3 seconds or less
• - Potential annual savings of $5 million through better off-specification load decisions
• - Reduced manual reinspection burden and weekly reporting effort
• - Faster, more transparent feedback for growers on inbound quality

Why It Matters

The SMBSC engagement shows how AI becomes operationally valuable when it is built around the realities of the workflow: environmental variation, high-volume seasonal processing, quality standards, alerts, and business decisions.

For SMBSC, the result was not just an AI model. It was a production workflow that helped standardize inspection, improve throughput, and create a stronger data foundation for future agentic AI work.