Intel VCaaS
AI Vision Checkout · Kenpath as technology partner
Fruits and vegetables checkout at retail stores averaged 35–40 minutes per customer. Cashiers manually identified every item, causing long queues and high labour costs. Intel needed a production-grade AI vision checkout system deployable on edge devices as the core of a scalable retail SaaS offering.
The slowest aisle in retail
Fresh produce checkout was the most painful bottleneck in retail — every item had to be manually identified by cashiers, there were no barcodes to scan. This wasn't just an inconvenience; it drove real costs in labour, queue abandonment and customer dissatisfaction. Intel needed a system that could work in real time on minimal in-store hardware, not a cloud-dependent prototype.
Computer vision that runs at the edge
We built an edge-deployed computer vision system that identifies fresh produce in real time, integrated with live pricing for seamless automatic checkout — no cashier identification required. A scalable ML-Ops pipeline with automated model retraining on edge devices, plus an OpenTelemetry observability layer, gives managers real-time operational analytics. We worked closely with Intel to co-develop the Edge AI framework, fine-tune vision models for produce recognition, and achieve real-time inference on minimal retail hardware.
The result
Solved a long-unsolved problem in retail — dramatically reducing checkout time and delivering the best checkout experience in the fruits and vegetables category.
System successfully deployed Edge AI, Advanced ML-Ops and real-time telemetry observability — all running on minimal in-store infrastructure.
“They came in with reputation of being part of the core team responsible for development and population scale deployment of various India stack based offerings. They underlined their expertise by developing a 1st time right cloud stack to be used as part of our SaaS offering. They were valuable partners and advisors on a host of other technology areas including AI.”
Raghavendra Bhat
Principal Engineer·Intel