Ximilar

Ximilar's visual AI platform transforms image processing through secure, scalable solutions for e-commerce, fashion, and manufacturing. Our technical breakdown reveals how this Czech-based startup handles billions of images annually while maintaining GDPR compliance and offering customizable deployment options.

Company Overview and Founders

The platform operates on a subscription-based model where users purchase API credits to process images. Each operation—such as tagging, recognizing objects, or performing image classification—is assigned a value in API credits. Users can choose from three tiered plans based on their monthly credit supply, with the basic plan offering 3,000 free credits.

Ximilar securely stores all customer data on their own EU servers, protected by multiple ISO standards. The company maintains intellectual property rights for generated models, allowing customers to retain ownership of their data and models. This arrangement is supported by positive feedback from clients like Babyndex and Filiberto Sosa of Sizes and Colors, who have reported successful integrations and improved system performance.

The company prioritizes data security through encryption at rest and in transit, role-based access control, and compliant systems that adhere to GDPR regulations. Their technical infrastructure includes certified hardware and multiple security certifications, ensuring reliable image processing across their diverse customer base. The platform's open architecture supports deployment on web applications, mobile devices, and edge computing systems, making it versatile for various use cases.

Core Technology and Capabilities

Ximilar's platform combines advanced image recognition and visual search capabilities through a modular structure that can be continually upgraded and enhanced. The technology supports multiple deployment options, including web applications, mobile apps, and offline functionality on edge devices, making it adaptable for various use cases.

The company's Computer Vision Platform serves multiple industries including e-commerce, fashion, retail, logistics, and security. Applications range from product counting in surveillance systems to recognition of collectible items, with automated grading systems for collectible cards. In the fashion industry, the platform enables personalized shopping experiences and improved product recommendations.

The technology processes each image once during synchronization, extracting descriptive data and immediately discarding it to handle large datasets without cloud storage of physical images. This approach allows processing of millions of images while maintaining privacy and security standards. The platform supports various image formats including jpg, jpeg, png, and tiff, with gif processing limited to the first frame.

Businesses can deploy the platform as customized solutions integrated into their existing workflows, with options for both off-premises cloud services and On-Premises installation for larger projects. The system scales to process over 15 billion images annually across multiple countries while maintaining GDPR compliance through multi-ISO certified data centers located in Prague.

The technical foundation of Ximilar's platform enables rapid model development through a combination of pre-trained AI modules and custom solution crafting. Users benefit from continuous model updates through the company's robust feedback mechanism, with improvements implemented without additional cost to maintain industry relevance.

Industry Applications and Customers

Ximilar processes over 1 billion images annually across 40+ countries, serving industries including Medtech, e-commerce, fashion, retail, logistics, and security. Their technology has generated a 380% increase in click rates for e-commerce clients and has been particularly effective in industries like fashion and retail.

Industry-Specific Applications

E-commerce

The platform drives significant improvements in product recommendations and the shopping experience. Notable clients like sizesandcolors.com and Profimedia report successful implementations across their vast image collections. The system has processed millions of images while maintaining data privacy under GDPR regulations.

Fashion and Retail

Key applications include automated grading systems for collectible cards and surveillance systems for personnel and product counting. Fashion applications span multiple industries, with capabilities including automatic tagging and image classification.

Manufacturing and Quality Control

The platform performs defect detection and surface inspection tasks, supporting both regular and edge device deployment. This capability has been operational for over 10 years at Profimedia, where it manages a collection of 100 million creative photos.

Technical Capabilities

The system handles various image formats including jpg, jpeg, png, webp, heic, bmp, tiff, and jfif, processing gifs using only the first frame. Each image is processed once during synchronization, with operations typically completing in 5-100 milliseconds per image for basic models, though performance varies based on input resolution and hardware.

Data Processing Workflow

Ximilar's architecture ensures minimal data retention, processing images in the cloud while immediately discarding descriptive data after extraction. This approach allows handling of large datasets without physical image storage, supporting up to 15 billion images annually while maintaining compliance with data protection regulations.

Customization and Integration

The platform offers both off-premises cloud services and on-premises installation options, with support for web applications, mobile apps, and edge device deployment. Customization capabilities include integration with existing workflows, training data annotation tools, and support for various deployment architectures.

Pricing Structure and Usage

Ximilar's pricing model operates on a straightforward API credit system, with each processing operation—such as tagging or recognizing objects—assigned a specific credit value. The company offers three tiered subscription plans based on monthly API credit supplies: Free, Business, and Professional. The Basic Plan provides 3,000 free credits per month, covering essential services like model training and basic image processing.

Users can maintain consistent service through API credit packs, which provide additional capacity beyond the monthly supply and do not expire. These packs are automatically activated after the monthly credit limit is reached, ensuring uninterrupted system performance during unexpected loads. All users have access to tracking features that notify them when 70%, 90%, and 100% of their credit supply is remaining, helping prevent service interruptions.

The company's architecture is designed to minimize costs while ensuring robust performance. Training custom image recognition models, deploying solutions, and making them accessible to end users are all free services, setting Ximilar apart from some competitors' pricing structures. Users pay only for actual service usage through API calls and credits, with no charges for idle time or training process costs.

Billing operates on a predictable monthly basis, with business and professional plan subscribers being billed on the same recurring day each month. When upgrading to a higher plan, users are billed at the upgraded rate from the same recurring day forward. Invoice emails are automatically generated and sent to users on their billing day, while the platform also provides access to invoices through user settings. Ximilar accepts payments through Stripe for secure credit or debit card transactions (Visa, MasterCard) and PayPal for both subscription and one-time purchases, with bank transfer payments accommodated through monthly invoices.

Data Security and Storage

All customer data is stored on secure servers located in the European Union, with multiple layers of protection including ISO certification. The company maintains strict access controls, allowing administrators to view images only when necessary and implementing role-based access management to limit data exposure. Customer intellectual property remains fully protected, with generated image models shared under a collaborative licensing arrangement that preserves user rights.

Data transmission between devices and servers uses end-to-end encryption, ensuring secure communication channels. The platform complies with GDPR regulations through multiple mechanisms, including regular security audits and compliance certifications. For clients with additional security requirements, the company offers flexible options including customized image access controls and data storage agreements.

The architecture supports both cloud-based and on-premises installations, giving businesses control over where their data is stored. All training data is managed on secure EU servers based on academic research standards, with time-limited access links used for specific project needs. The company's technical infrastructure includes certified hardware and multiple security certifications, providing a robust foundation for image processing tasks.