Smart Vision - An AI-native empowerment platform
Smart Vision is an AI-native platform that integrates GenAI into enterprises by combining computing power, models, knowledge, and applications. It streamlines the four key stages of GenAI engineering: selection, training, usage, and management, offering a comprehensive suite for digital transformation. With modules for model and computing power management, enterprise private knowledge management, and AI application engineering, Smart Vision provides advanced, enterprise-ready large models, optimizing computing power, integrated scheduling, and efficient training.
Unified Management of Heterogeneous Computing Power &Optimized Model Performance
Smart Vision supports unified management of diverse CPU and GPU resources across different environments, addressing large model computing bottlenecks and optimizing GPU usage based on model and chip types.
One-Click Access to Mainstream Models &Efficient Private Model Fine-Tuning
Smart Vision offers over 40 foundational models for flexible business applications. It includes a comprehensive toolset for private model fine-tuning, enabling businesses to customize high-accuracy models.
Multimodal,Multilevel Knowledge Management for Data Asset Activation
Smart Vision integrates various data formats—text, images, video—with access controls to ensure data security, creating an enterprise-grade private knowledge base.
Flexible &Ready-to-Use Application Building
Smart Vision supports no-code, low-code, and full-code configurations across five development modes, seamlessly integrating with corporate systems and offering pre-built templates for efficient application development.
 
 
 
Applications
Computational Deployment
Model Training
Data Governance
Knowledge Q&A
Copywriting
Document Reading
Document Parsing
Content Translation
Text-to-Image Generation
Super Employee
Smart Customer Service
 
 
 
 
 
Computational Deployment
Supports dynamic scheduling across multi-brand heterogeneous computing power, including domestic, non-domestic, public cloud, and private cloud setups, with custom development services to meet diverse enterprise needs.
Model Training
Offers no-code model training configuration, supporting inference acceleration technologies like accelerated engines, model quantization, and compression, helping reduce latency, save resources, and enhance model stability and reliability.
Data Governance
Unlocks the full potential of existing big data platforms by addressing limitations in processing unstructured data, providing comprehensive data governance to maximize value.
Knowledge Q&A
Delivers precise answers to business questions through natural language interaction with a private large model trained on enterprise-specific knowledge.
Copywriting
Features built-in templates for enterprise-grade copywriting, allowing users to generate high-quality content by simply filling in the necessary information.
Document Reading
Supports one-click document summary generation, enabling in-depth reading and interactive questioning of document content.
Document Parsing
Allows batch processing of documents into specified formats, facilitating the creation of data assets.
Content Translation
Provides multi-language translation and correction services, aiding enterprises in expanding globally.
Text-to-Image Generation
Automatically generates high-quality images from natural language descriptions, with support for professional image parameters, enabling creative transformation from text to visual art.
Super Employee
Efficiently queries internal enterprise knowledge through natural language, offering 24/7 online support and autonomous learning from accumulated knowledge.
Smart Customer Service
Provides around-the-clock consultation, processing, and inquiry services via natural language interaction, delivering instant responses to user needs, improving service efficiency and satisfaction while reducing the burden on human customer service.
Use Cases
Intelligent Computing Infrastructure
Shenzhou Kuntai AI Server
KunTai A722 Inference AI Server
A next-generation ARM-based AI computing server powered by the Huawei Kunpeng 920 processor and Ascend Atlas AI accelerator card. It features exceptional computing power, full-scenario AI capabilities, and an open ecosystem. The server is designed for image recognition, video processing, inference computing, and machine learning, and is widely used in smart city surveillance, intelligent transportation, smart healthcare, and AI inference.
  • 2
    Kunpeng 920 processors
  • 16or32
    DDR4 RDIMM
  • 2933MT/s
    memory speed
  • Supports up8cards
    Atlas 300V Video Analysis Card
    Atlas 300I Pro Inference Card
    Atlas 300V Pro Video Analysis Card
  • 1120TOPS INT8
    Up to
Integrated GenAI Suite
Shenzhou Kuntai & Smart Vision Integrated GenAI Suite
Integrating Smart Vision with our proprietary server brand Shenzhou Kuntai, we provide a holistic solution for agile deployment of private-domain GenAI. This solution addresses enterprise-level clients' needs for agile deployment and security in private-domain deployment of GenAI. It supports inference for multiple large models and a wide range of AI model applications, helping clients customize exclusive AI models using their own data to drive business innovation.
  • Powerful Performance
    Supports large models with up to 200 billion parameters.
  • Ready to Use
    Integrated hardware and software, providing private domain computing power that's ready out of the box.
  • Custom Solutions
    Supports customization of enterprise-specific large models with a private knowledge base.
  • One-Click Deployment
    Built-in platform with integrated features for easy deployment.
  • One-Stop Construction
    Offers a complete suite of full-stack AI services, easy access to computing power, models, and applications.
AI Computing Power Scheduling & Acceleration Platform
Heterogeneous Intelligence Schedule Operation Platform ——HISO
Supports essential features such as mixed networking, hybrid scheduling, and fine-grained computing power isolation for domestic and international GPUs. HISO enables easy management of multiple GPU clusters as if they were a single host, optimizing resource utilization by matching the best computing power combination from the entire intelligent computing resource pool based on user needs.
Heterogeneous IIntelligence Computing Acceleration Platform ——HICA
HICA masks the differences in underlying computing power ecosystems within clusters, breaks through key computational efficiency bottlenecks, and enhances computing power utilization and availability.
 
Computing Power and Services for AI Computing Centers
  • Shenyang AICC
    Powered by Shenzhou Kuntai's advanced computing products, the Shenyang Artificial Intelligence Computing Center is a high-performance, energy-efficient facility designed to be the fastest AI training cluster in Northeast China. It provides accessible AI computing power, reduces costs, and attracts AI companies to the Hunnan District, supporting regional AI development.
  • Xiamen Kunpeng Supercomputing Center
    The supercomputing center delivers reliable computing and storage for government, enterprises, universities, and research institutions. It supports digital governance, smart healthcare, smart industry, smart education, and intelligent manufacturing through supercomputing, AI, data centers, and big data. By leveraging heterogeneous integration, it provides a seamless experience in storage, computing, and operations, driving innovation in intelligent infrastructure.
  • Shenzhen AICC
    Constructed to national Class A and international Uptime Tier III standards, this computing center offers a one-stop solution for high-end computing needs through its AI computing platform. It is suitable for various applications including scientific research, deep learning, data analysis, and enterprise resource planning. The center provides reliable, efficient, and secure computing services to its users.