KoiReader launches Physical AI platform for yard automation

Full-stack yard automation goes live at enterprise scale

KoiReader Technologies has released its KoiVision Yard Intelligence Suite for general availability, marking a significant step in automating one of the supply chain's most manual bottlenecks. The platform combines proprietary edge hardware with computer vision models to automate gate entry, track assets across yards, and coordinate dock operations—without requiring RFID tags, Bluetooth beacons, or drone infrastructure.

The system is already deployed across production facilities spanning more than seven million square feet, including Fortune 50 customers, according to details first reported by Automation Watch. The platform handles capacity for 3,000 trailer and container positions and integrates with existing transportation management systems, yard management systems, and ERP platforms.

Why it matters

U.S. detention costs from yard inefficiencies exceed $15 billion annually, according to the American Transportation Research Institute. While warehouse interiors have seen waves of automation investment, yards have remained largely paper-based and disconnected. KoiVision addresses this gap with a unified platform that works across verticals and geographies from day one, rather than requiring site-by-site customization—a critical advantage for enterprises operating multiple facilities.

Nine products on a single AI backbone

The suite comprises nine integrated products spanning the full yard workflow. KoiDriver and KoiKiosk handle driver check-in through mobile apps and AI-powered kiosks that validate documents against backend systems. KoiGuard automates gate processing with computer vision, capturing data and managing multi-lane traffic while integrating with weigh scales.

KoiRover mounts vision AI hardware on yard mules to track trailers and containers to slot-level precision through continuous scanning. KoiTwin creates digital twins of facilities using LiDAR scans, while KoiAssist orchestrates yard-mule movement to increase scan frequency. KoiConnect handles integration with enterprise systems across the full lifecycle from purchase orders to invoice settlement.

KoiGPT, a domain-specific large language model, provides natural-language access to yard efficiency analysis. An upcoming product, KoiPass, will authenticate trucks in 15 seconds using telemetry, voice, and vision AI.

Trained across industries and geographies

KoiReader developed the platform in partnership with enterprise customers across automotive, consumer packaged goods, food and beverage, manufacturing, retail, intermodal operations, and logistics providers. As an NVIDIA partner, the company trains its vision models on a global dataset spanning the Americas, Europe, Latin America, and Asia-Pacific, augmented with synthetic data engineered to cover edge cases like extreme weather and unusual trailer configurations.

Founder and CEO Ashutosh Prasad, who left a vice president role at XPO Logistics to launch KoiReader in July 2019, emphasized the company's approach. "We treat it as a Physical AI problem—hardware and software engineered together, trained across verticals and geographies rather than one customer at a time," he said.

Compliance and telematics integration

For facilities operating under Foreign Trade Zone rules or Customs and Border Protection oversight, the suite automates CTPAT-mandated physical inspections using AI kiosk cameras and verifies high-security seals through the View-Verify-Touch-Tug method. The platform logs an immutable audit trail for CBP validation.

Telematics integration connects inbound vehicle GPS, OBD-II, and electronic logging device data with yard management systems ahead of arrival, enabling geo-fenced gate preparation and pre-entry checks on fuel, refrigeration temperature, and diagnostic codes across more than 100 telematics vendors.

KoiReader holds six granted U.S. patents in Physical AI, with three foundational to the Yard Intelligence Suite covering data extraction, yard mapping and asset tracking, and inventory monitoring. The company offers its ruggedized hardware through a Hardware-as-a-Service model built on NVIDIA, Intel, and AMD ecosystems.

Details of the launch were first reported by Automation Watch.