Introducing the world’s first

Retail AI Foundation Model

GET IN TOUCH
RETAIL WITHOUT LIMITS

OUR MISSION

We provide the world’s first Retail AI Foundation Model

We’re enabling the $32 Trillion retail industry with product and people recognition models that bring unmatched accuracy, speed, and scalability to market.

Our foundation models are designed to be efficiently deployed and run at the edge on low cost, low power, small form-factor hardware, and are protected by over 50 patents.

We provide the world’s first Retail AI Foundation Model

The Problem

We address the retail industry's $174B+ concern –
shrink and staffing

$112B

$112B

lost to external/internal shrink

$62.6B

$62.6B

spent on checkout staffing

Two massive problems, one root cause – a lack of reliable, scalable, product / people identification.

Current computer vision solutions cannot identify products and people accurately enough.
80% accuracy is a non-starter for established retailers.

Our Solution

We’ve pioneered the first commercially viable
AI Computer Vision solution for retail

“Commercially viable” means human-level or better product identification accuracy
with computer-level speed and infinite scalability.

Accuracy

Accuracy

99% or better identification

+20pp to +70pp better*
Speed

Speed

~50ms per product (20fps on low power edge device)

20x faster
Scalability

Scalability

SKU counts exceeding 250K (~2x the largest superstore)

100x more
Ease

Ease

Enroll 250K+ SKU product catalogs in minutes

400x cheaper
Resiliency

Resiliency

Functions with or without internet connectivity

100% uptime

* All comparisons based on 2024 Walmart competitive bake-off against a dozen leading competitors

Use Cases

Built for every aisle, every store, every moment

From self-checkout to stockrooms, Ultron AI sees what others miss - flagging shrink and fraud, validating returns, ensuring stocked shelves, and optimizing flow. It’s always on, always learning, and always built for what retail throws your way.

AI-Powered Product Identification

AI-Powered Product Identification

Instant recognition of 250K+ products with sub-50ms processing speed.

AI Checkout & Shrink Protection

AI Checkout & Shrink Protection

Seamless, barcode-free self-checkout with fraud and shrink detection.

Intelligent Shelf Monitoring

Intelligent Shelf Monitoring

Ensure planogram compliance and track real-time inventory movement.

Smart Cart Tracking System

Smart Cart Tracking System

AI-driven cart verification for frictionless self-checkout experiences.

AI-Powered Loss Prevention

AI-Powered Loss Prevention

Detects high-risk behaviors in real time to prevent theft and reduce shrink.

Augmented/Virtual Reality in Retail

Augmented/Virtual Reality in Retail

Immersive shopping experiences with AI-driven augmented reality.

Our Partners

walmart-2 fujitsu-2 elo-2 hailo-2 hp-2 egen-2 datalogic hanshow nvidia-2 525k clarity

Product Videos

Product in action

Watch how Ultron AI detects fraud, verifies returns, and powers smarter stores. Real footage, real speed, real results. No simulations. Just retail intelligence at work.

View all videos
Home

UltronAI Self Checkout

Our AI running on a Hailo-8 AI accelerator on an HP checkout system

Home

Live Ar Rasberry Hailo8

UltronAI Augmented Reality Glasses:driving real-time Augmented Reality to help store operations

Home

Ultron AI Shrink Prevention

Ticket Switching Detection on HP Engage POS w/ Hailo AI Accerelator

Home

UltronAI Cart Receipt Validation

UltronAI verifies products bought by simply looking in the cart

Home

UltronAI Self-Checkout on Fujitsu system

Checkout our UltronAI Visual AI solution integrated with a Fujitsu camera system identifying very similar looking products running on an nVidia Jetson Nano.

Home

UltronAI Shelf Inventory Analytics

Using our AI technology to perform shelf inventory for out of stock and identify misplaced items

Home

UltronAI Visual AI Self-Checkout

UltronAI technology using zero-shot enrollment gallery of 250k SKUs running on an nVidia Jetson. 

Luke Wilwerding 

This innovation is a powerful step forward in reducing shrink and the elevating customer experience. And it’s just the beginning of our AI journey in collaboration.

Luke Wilwerding 

Luke Wilwerding 

Vice President, North America Sales @ Elo Touch Solutions

Greg Cathey

The work happening here is truly game-changing,
and it was inspiring to collaborate on solutions
that could redefine the future of retail.

Greg Cathey

Greg Cathey

Senior Vice President, Transformation & Innovation @ Walmart

Tracy Flynn

From retail product recognition to advanced surveillance, it’s exciting to see robust AI use cases delivered in compact, power-efficient platforms.

Tracy Flynn

Tracy Flynn

North America OEM Business Development @ HP

How It Works

We integrate with any system and are hardware agnostic
to power any retail CV use-case

Any Point of Capture

Any Point of Capture

Camera, phone, AR/smart glasses, image file, X-ray

Any Processor

Any Processor

Running on any edge device, or any cloud service for mobile

Any Object

Any Object

Unlimited ID of SKUs, faces, actions & events - via single-shot enrollment

Any Integration

Any Integration

SDKs and APIs to deliver Ultron through any OEM

Any Environment

Any Environment

Indoor, outdoor, high or low-traffic store layouts

Any Use Case

Any Use Case

Asset protection, checkout, inventory, analytics, simulation

Intellectual Property

Our Patent Portfolio

Ultron AI owns the foundational IP behind retail-grade computer vision. That means every deployment is protected, every integration is covered, and every rollout reduces your exposure – not adds to it.

View full portfolio
#US11900516B2
System And Method For Pose Tolerant Feature Extraction Using Generated Pose-Altered Images
+
This system improves product recognition by generating synthetic 2D views from a single image using 3D modeling, allowing neural networks to learn pose-invariant features. These features are stored in a database for classification and matching, helping overcome data limitations in training AI models.
#WO2023192666A1
System and Method for Multiview Product Detection and Recognition
+
A multi-camera system that detects and identifies products from any angle by extracting and fusing visual features, enabling accurate recognition even when items are stacked, obscured, or randomly placed—ideal for automated checkout and complex retail environments.
#US12217339B2
Multiple Hypothesis Transformation Matching For Robust Verification Of Object Identification
+
A system that enhances product recognition on retail shelves by generating synthetic 2D views from pristine images, training AI models to recognize items from any angle or condition—boosting accuracy in inventory tracking, planogram compliance, and shelf management even with limited real-world data.
#US12067527B2
System and Method for Identifying Misplaced Products in a Shelf Management System
+
An autonomous inventory monitoring system using a mobile robot equipped with multiple cameras to scan store shelves, identify products, and detect out-of-stock or misplaced items—eliminating the need for manual planograms and enabling real-time shelf compliance and inventory accuracy.
#US10430922B2
Methods And Software For Generating A Derived 3D Object Model From A Single 2D Image
+
A method and software system for generating accurate 3D object models from a single 2D image by using landmark detection, projection estimation, and model warping—enabling applications like pose-invariant face recognition and 3D reconstruction without the need for multiple image views.
#US12131260B2
Discriminative Cosine Embedding in Machine Learning
+
A novel loss function—Copernican Loss—that enhances deep learning by minimizing intra-class variation and maximizing inter-class separation using cosine distance. It improves training-test alignment, requires no hard sample mining, and achieves efficient, highly discriminative embeddings using simple mini-batch statistics.

Want to learn more?

Let’s talk about how Ultron AI can reduce costs, speed up operations, and increase your bottom line - all while making magical customer experiences.