Smart Vision - Enterprise-level Agent Middle 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
Super Employee
Intelligent Document Review
AI for BI
Data Asset Management
Data Governance
Model Training
 
 
 
 
 
Super Employee
It includes knowledge Q&A, data Q&A, and intelligent process module. Equipped with the intent recognition capability of LLM, it efficiently queries internal enterprise knowledge and provides 24/7 online services for employees. Through natural language interaction, combined with the application of private LLMs trained in the enterprise's private domain knowledge base, it can understand, remember and reason about business problems, and comprehensively and accurately answer users' questions.
Intelligent Document Review
Through intelligent document generation, review and knowledge base management, it examines whether documents comply with laws, regulations and industry standards, thereby enhancing the level of enterprise process automation, improving work efficiency and strengthening compliance.
AI for BI
By equipping BI with natural language processing technology, enterprise decision-makers only need to state their questions to the computer to obtain answers, making business intelligence widely accessible and helping enterprises establish a data-centered and metric governance management model.
Data Asset Management
By getting the data from a comprehensive perspective, it is easy for enterprises to find, understand, recommend, apply for and evaluate their data, helps enterprises establish and manage data assets, explore the value of data assets and drive business innovation.
Data Governance
Many enterprises have accumulated a large amount of unstructured data internally, but due to cost and efficiency issues, they have never been able to fully utilize it. With the help of Smart Vision, comprehensive data governance can be achieved, and the value of the enterprise's big data platform can be unlocked.
Model Training
It provides zero-code model training configuration methods and supports inference acceleration technologies such as acceleration engines, model quantization, and model compression, which can help enterprises reduce inference latency, save resources, and enhance the stability and reliability of LLMs.
Smart Vision——Intelligent Process Workbench
Centered on process-driven, closely aligned with enterprise business scenarios, it provides clear task guidance for business personnel; seamlessly integrates all business application capabilities to achieve efficient data integration and sharing; deeply integrates AI Agent capabilities to comprehensively empower business scenarios, helping enterprises improve efficiency, enhance quality, reduce costs, and promote the intelligent upgrade of business processes.
Business Process Management
Avoiding differences between process design and actual execution
• Visualize business processes in the form of an operation map, clearly showing each step, the relationships between nodes, and the current stage of the process
• Analyze task processing time, resource consumption and other indicator data using AI, identify bottlenecks and provide process optimization suggestions
• Empower process managers to quickly define business processes without coding—using drag-and-drop and parameter configuration.
Business Application Integration
Achieving data asset accumulation and data insights
• By connecting CRM, ERP, OA and other business systems through standardized interface protocols, real-time data transmission and sharing are realized, ensuring consistency of business data
• Building a unified data model, through in-depth data mining and integration using data analysis tools, business reports and analysis reports are generated to provide data support for business decision-making
• According to user roles and responsibilities, data access permissions are assigned to ensure data security and prevent the leakage of sensitive information
AI Agent Quick Integration
Empowering Business, Rapid Iteration
• Quickly integrate the adapted AI capabilities for integration, providing AI assistance for business processing
• Deeply combine AI capabilities with business processes, and apply them in scenarios such as customer service, sales forecasting, and risk assessment
 
 
 
Smart Vision Deepseek Edition
Smart Vision Deepseek Edition adopts four core strategies - private deployment, compatibility with information innovation, scenario empowerment, and integrated solutions - to address the four major core pain points that enterprises encounter during the deployment of DeepSeek: data security anxiety, the dilemma of information innovation compatibility, the lack of application scenarios, and the high costs of trial and error and computing power. By precisely resolving these deployment pain points, it helps enterprises deeply integrate AI technology into their core businesses, truly achieving cost reduction and efficiency improvement, then to become a key force in breaking through in the intelligent era.
Private Deployment
Tailored AI solutions for enterprises
• On-demand planning for full-stack deployment, from DeepSeekR1 14B to the fully-powered DeepSeek version
• Smart Vision Deepseek Edition includes a compute optimization module, reducing the computing cost by 30%
• Model selection service to make the deployment plan more suitable for the actual situation of enterprises
• Backed by 20-year of IT service expertise---proficient in deployment and also knowledgeable about enterprise-level services.
Innovation-Driven Adaptation
Seamless integration with diversified ecosystems
• KunTai Smart Vision all-in-one appliance (DeepSeek Edition) for secure and rapid deployment
• Full-stack private deployment solution + full-stack independently controllable innovation-driven services adaptation
• Deep integration of software and hardware, providing an integrated solution
• Private deployment capabilities covering industry scenarios
Professional Training Suite
Building enterprise-level exclusive models
• Data distillation technology enables high-quality data purification, providing reliable data for model training
• Compliance-Scenario Adaptability - Performance Index Three-level Evaluation System ensures high-quality and efficient model training
• Instruction fine-tuning training / Pure reinforcement learning / Multi-stage hybrid training methods makes models better aligned with enterprise needs.
Scenario Empowerment
Making AI Technology Truly Operational
• Smart Vision(DeepSeek Edition) offers solutions across eight major industries
• Covering digital employees, intelligent customer service, intelligent procurement and other scenarios
• Helping enterprises quickly identify the entry point for AI implementation, allowing the technology to truly serve the business
• Achieving deep empowerment for “AI for Process”
 
 
 
AI Smart Vision
AI Smart Vision is dedicated to the principle of "Bring AI to Data, not Data to AI". Through localized and lightweight design, it enables enterprise-level AI capabilities to be deployed to individual PC terminals. Whether it's developers, enterprise employees, or ordinary users, they can build their own exclusive LLMs locally through low-cost deployment, free model switching, and the construction of private knowledge bases.
One-click deployment of local LLMs
The first step of technological democratization
By using model distillation and optimization, the parameters of AI Smart Vision are compressed to a lightweight version that can be run locally. Users only need to perform a one-click deployment, and they can complete the installation of the local model on their personal PC.
Free Switching Between Local Multi-Model
Flexible response to diverse demands
Through free switching between local multi-model, users can call the most suitable model as needed, without being limited by the capabilities of a single model. At the same time, all models can run offline and do not rely on cloud computing power.
Personal Private AI Knowledge Base
Achieving both data security and intelligent management
The platform supports one-click import of personal documents, notes, images, etc. Through the Embedding model, it automatically completes content parsing, tag classification and vectorized storage, forming a structured knowledge network. Whether it's personal documents, knowledge notes, or enterprise sensitive information, users can build their own AI knowledge base through local private storage and intelligent file management functions.
Plug-and-Play
Unlock the Full Potential of Local Large Models
AI Smart Vision integrates mainstream AI tools and application templates both at home and abroad. Users can use it out of the box without any coding background. It covers the entire process from creation to management: AI-assisted writing, intelligent synchronization, flexible invocation, precise polishing, and ask AI to answer users' questions at any time.
 
 
 
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
KunTai Smart Vision Integrated Machine
KunTai Smart Vision Integrated Machine
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.


Shenzhou Kuntai Smart Vision All-in-One Machine (DeepSeek Edition)
Through a deeply optimized private deployment solution, the Shenzhou Kuntai Smart Vision All-in-One Appliance(DeepSeek Edition) provides enterprises with efficient and stable AI services. Through professional training packages, it helps enterprises create exclusive R1 models, ensuring that the model performance is highly matched with the enterprise's needs. It offers eight industry solutions to help enterprises quickly identify the entry points for the implementation of AI, assist them in deeply integrating AI technology into their core business, and truly achieve cost reduction and efficiency improvement.
  • Secure deployment
    Simplify deployment and lower technical threshold
  • Full-stack service
    Full-stack private deployment solution + full-stack information technology innovation service adaptation
  • Deep integration
    Deep integration of software and hardware, providing an integrated solution
  • Deeply cultivating the industry
    Private deployment capabilities covering industry scenarios
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.