Integrated Quality, Equipment, and Safety Management Solution Based on the DX-AI Platform
Smart Manufacturing Innovation Utilizing Domain AI Agents and Ontology-based RAG Technology
Introducing an AI solution that revolutionarily integrates quality, equipment, and safety management in the manufacturing field.
By combining domain expertise and the latest AI technology, we implement a smart manufacturing environment to simultaneously achieve productivity improvement and safety management.
Index
1. Recurring Problems in Manufacturing Facilities
2. Domain-Specialized AI Agents Based on the DX-AI Platform
3. Key Technology Architecture of the DX-AI Platform
4. DX-AI Platform Architecture Blueprint
5. Service Scenarios
6. Summary of Key Features
7. Detailed Architecture and Technology Stack
8. Effectiveness and Scalability for Job Roles in the Field
1. Recurring Problems in the Manufacturing Facility
Dispersed Information
Equipment/quality/safety information is scattered across different systems, making it difficult to manage in an integrated manner.
Lack of Interconnected Analysis
It is difficult to identify the linkage between equipment failure, quality issues, and safety incidents.
Loss of Expertise
Veteran's field expertise is not documented, leading to risks when personnel leave the organization.
Difficulty in Compliance
It is challenging to comply with complex regulations, increasing the possibility of legal issues.
These problems reduce the efficiency of the manufacturing facility and increase safety risks. An integrated intelligent system is needed to address these issues.
2. Domain-Specialized AI Agent Based on the DX-AI Platform
Integrated Data Collection
Integrates generative AI, RAG, ontology, IoT, and voice recording into a single system.
Real-Time Data Understanding
AI understands documents, drawings, regulations, and voice data in real-time.
Risk Prediction and Response
Performs risk prediction, provides action recommendations to workers, and automates record-keeping.
Field Partner
Not just a chatbot, but an integrated partner that understands equipment, quality, and safety.
The DX-AI Platform integrates and understands the complex data in manufacturing environments to provide real problem-solving and decision-making support. It provides the necessary information to all stakeholders, from frontline workers to managers.
3. DX-AI Platform Core Technology Architecture
1
Data Ingestion & Integration
  • Source Data: IoT sensors, PLC, MES, ERP, QMS, drawings, SOPs, regulatory documents, voice recordings, images
  • Processing Technologies:
  • Data Pipeline / IoT Prism
  • Document Parsing (OCR, STT, Rule-based Parsing)
  • Image Processing & OCR
  • Data Consistency Management (Noise Removal, Resampling, Labeling)
2
Pre-Processing & Feature Engineering
  • Ontology & Knowledge Graph Construction (RDF, Neo4j)
  • Feature Labeling (Contextual Feature Generation)
  • Data Catalog & Schema Registry (Data Standardization Management)
3
Core AI & ML Service Layer
  • ML Models:
  • Regression & Classification (CatBoost, XGBoost, etc.)
  • Time-series Forecasting (Prediction)
  • Clustering & Unsupervised Learning
  • AI Engines:
  • LangChain + FAISS (RAG Framework)
  • LLM (Gemma3, Ollama, QLoRA Fine-tuning)
  • Inference & Analytics:
  • RAG, Ontology, Embedding, Semantic Search
  • XAI (Explainable AI)
4
Domain-Specific Engine
  • Quality Engine (Prediction & Simulation)
  • Anomaly Detection Engine (Fault Prediction)
  • Maintenance & Compliance Engine (CMMS, Regulatory Analysis)
  • Planning & Scheduling Engine
5
GenAI Intelligence Layer
  • Prompt Orchestration & Template Management
  • Fine-tuning & Customization (QLoRA, PEFT-based)
  • Conversational Agent & Domain AI Agent
  • Multimodal Support (Text, Image, Speech)
6
Output & UX Layer
  • Conversational Interface:
  • Conversational Agent (Chatbot UI)
  • Natural Language Query (Text/Voice Q&A)
  • Visualization & Notification:
  • Interactive Dashboard (Status/Monitoring)
  • Report Generator (EDS/SEM Analysis Reports)
  • Notification Service (SMS, Push Notifications)
  • API & Export:
  • REST/gRPC API
  • ERP/MES Integration
  • PDF/HTML Report Generation
7
Cross-Cutting Concerns
  • Security & Governance (Data Security, Air-gapped Network)
  • MLOps & Model Monitoring (Performance Monitoring, Retraining)
  • Configuration & Personalization (User Customization)
  • Metadata & Audit Log Management
4. DX-AI Platform Architecture Blueprint
5. Service Scenario 01
DX-AI IoT Prism
An integrated framework for effectively processing, analyzing, and visualizing data collected in a manufacturing IoT environment.
It helps non-experts easily utilize the value of IoT data through AI-based data analysis, automated analysis curation, and question-answering capabilities using generative AI.
DX-AI Modeler
Supports data scientists' decision-making by automatically performing ML modeling and performance comparison.
Maximizes model performance through hyperparameter optimization.
DX-AI Data Analyzer
AI agents analyze and visualize various data.
It understands the user's analysis requirements, explains the results, and recommends analysis approaches including visualization.
To be updated
DX-AI iDSB
An intelligent dashboard (iDSB) system that integrates quality, equipment, and safety management in the manufacturing field.
DX-AI iDSB is an AI dashboard that helps predict and manage quality and equipment issues in the factory, allowing you to predict equipment status or quality risks without inspection and quickly identify and respond to the root causes.
5. Service Scenario 02
DX-AI Advisor
Extracts various document information based on ontology and generates expert-level reports on specific topics.
DX-AI Occupational Safety and Health Act Chatbot
Searches and retrieves the contents of the Occupational Safety and Health Act based on ontology.
Understands various industrial situations and searches for relevant Occupational Safety and Health Act provisions to provide information.
DX-AI Fire Services Act Chatbot
Searches and retrieves the contents of the Fire Services Act based on ontology.
Provides accurate answers through contextual and associative analysis to explore information.
5. Service Scenario 03
DX-AI IoT CMMS
It recommends necessary actions based on real-time IoT data-based alarms and past CMMS information.
5. Service Scenario 04
Dynamic Production Planning
Automatically establishes and optimizes manufacturing production plans.
Reflects real-time plans considering production line productivity, and provides AI analysis and responses to questions about the production plan.
6. Summary of Key Features
RAG-based Conversational AI Agent
Based on various documents, drawings, and data sources, you can query in natural language for information needed on the field, and it will provide immediate and accurate answers with supporting materials.
Drawing OCR and Image Analysis
It automatically recognizes and extracts locations, components, and work points from equipment drawings and images, which can be utilized for analysis and work support.
Ontology-based Reasoning
By building a semantic knowledge graph that connects equipment, quality, and safety data, it supports risk assessment, regulation judgment, and relevance inference.
Voice Recording Conversion
It preserves field knowledge through the process of voice → text → vector storage.
Real-time IoT Monitoring and Alerts
It collects and analyzes sensor data in real-time, detects anomalies, and provides warning alerts and action recommendations.
Quality and Equipment Predictive Analytics
Through AI-based quality prediction and equipment failure analysis, it supports proactive response and optimized maintenance.
Integrated Dashboard and Reporting
The dashboard that allows you to see quality, equipment, and safety data at a glance, and the automatically generated reports enable quick decision-making and sharing.
These functions are organically connected to provide integrated support for quality, equipment, and safety management in the manufacturing field. Each function has value on its own, but when integrated, it can achieve greater synergy.
6. Real-time Monitoring and Analysis
Real-time Monitoring Alerts
It provides real-time monitoring of equipment status and immediate alerts for abnormal conditions.
Operators can quickly grasp the current situation through the dashboard.
Drawing OCR Results
It automatically extracts and digitizes important information from complex equipment drawings.
This makes it easier to search and analyze drawings, and quickly find relevant information.
Semantic Analysis
It visualizes the relationships and impact between equipment using an ontology-based approach.
This allows intuitive understanding of the complex associations.
The demo screens of the DX-AI Platform are designed with a user-friendly interface to easily comprehend complex data.
It provides a variety of functions from real-time monitoring to in-depth analysis.
6. Intelligent Conversation and Analytics System
RAG Chatbot Response
The document-based question-answering system immediately provides the information needed on-site. Without having to search through complex manuals or SOPs, you can ask questions in natural language and receive accurate answers.
Lookup of Regulations/Laws
It provides accurate answers and sources for questions about safety regulations or legal requirements. This allows you to easily verify compliance and reduce legal risks.
Predictive Model and Result Interpretation
It interprets the results of AI-based predictive models in an easy-to-understand format. This allows field workers to easily understand and utilize the results of complex data analysis.
The DX-AI Platform goes beyond just providing information, it analyzes and interprets data to provide the insights needed for decision-making. This enables rapid and accurate responses in the field.
7. Detailed Architecture Description - Data Sources
Equipment Diagrams
We collect and analyze equipment diagrams in various formats such as PDF/Image.
Site SOP Documents, Regulations/Laws
We collect standard operating procedure documents and related regulations in formats such as PDF/Text.
IoT Sensor Data
We collect real-time IoT sensor data that monitors the status of the equipment.
Voice Recording Data
We utilize STT (Speech-to-Text) technology to convert on-site voice recordings into text.
We efficiently collect and store various types of data using the MISO Multi-modal BIG DATA platform. In this step, we secure the data sources and establish the foundation for subsequent processing.
7. Detailed Architecture Description - Data Processing
We utilize image OCR and text extraction solutions, Text Analysis solutions, and STT conversion engines to convert various data formats into standardized forms. This process prepares unstructured data into a format suitable for analysis.
Document Text Extraction and Classification
Extract text from documents and automatically classify them based on metadata.
IoT Sensor Data Preprocessing
Standardize sensor data in real-time and process it into an analysis-ready format.
Voice Data STT Conversion
Convert field audio recordings into text format for searchable storage.
Blueprint Image OCR Processing
Automatically recognize and extract text and components from facility blueprints.
7. Detailed Architecture Description - Ontology and Vector Data Storage
Ontology DB
Using RDFLib and Neo4j, we build semantic relationships between equipment, quality, and safety. This allows us to go beyond simple keyword searches and understand the associations between concepts.
  • Define relationships between equipment
  • Map connections between failures, quality, and safety
  • Link to regulations and laws
Vector DB
Utilizing FAISS, we construct a vector index for real-time Retrieval-Augmented Generation (RAG) search. This enables accurate information retrieval for natural language queries.
  • Store document embeddings
  • Similarity-based search
  • Real-time updates
By leveraging technologies like RDFLib, Neo4j, FAISS, and MISO Object Storage, we efficiently store and manage the semantic relationships and vector embeddings of the data. This stage forms the foundation for accurate AI inference and search.
7. Local Generative AI Operating Environment
MISO AI Model Management and Operation Platform
Supports model deployment, monitoring, optimization, and retraining
RAG Engine
LangChain-based data retrieval and inference
Multimodal Data Platform
Data Catalog, Vector DB
LLM/sLLM Models
Based on Gemma3, Ollama
The DX-AI Platform operates generative AI within Miso Information Technology's local and closed network environment without external cloud, protecting the customer's sensitive data and providing immediate response speed. It enables local/closed network deployment and operation of generative AI, complete removal of cloud dependency when processing sensitive information, and real-time ontology-based intelligent response generation.
7. User Interface
Real-time AI Chatbot Service
This is a conversational interface that provides relevant information and action plans immediately in response to natural language queries. You can quickly find the information you need on the field.
Operations and Management Dashboard
This is an integrated dashboard that allows you to see equipment status, quality metrics, safety status, and more at a glance. Real-time monitoring and trend analysis are possible.
Warning Alert System Based on Risks and Status
This is a warning system that provides immediate alerts and response plans when abnormal conditions occur. It supports accident prevention and rapid response.
We provide a user-friendly interface by utilizing the MISO Smart VI (Vision Insight Platform) and AI-based chatbot solutions. The intuitive UI/UX allows you to easily leverage complex data and AI capabilities.
7. Evolving AI in the Field by Following Feasible Steps
Step 1: Prompt-based Q&A
We will design prompt templates and develop a prototype domain document Q&A system. We will load quality manuals, SOPs, and checklists in PDF format and answer questions like "What was this criteria?", "What are the precautions during the operation?"
Step 2: Building a RAG System
We will implement a document search based on LangChain and strengthen the search results using FAISS + ontology. Integrated search of regulations, drawings, and reports is possible with real-time updates and continuous improvement.
Step 3: Designing a Domain AI Agent
We will implement a chatbot based on user situation understanding and propose actions based on the context of the question. It will automatically respond to "What should I do when this equipment fails?" and include explanations based on PHM results.
Step 4: Fine-Tuning a Dedicated LLM
We will train on quality, PHM, and regulation data, and develop a lightweight model based on QLoRA. We will learn the customer-specific response patterns and be able to run independently in an on-premise environment.
Instead of a large-scale deployment without experience, the goal is to start small and evolve the AI with customer data. Each step is based on the achievements of the previous step, and is continuously improved by reflecting the feedback from the actual field.
7. Technology Stack
LangChain + FAISS
Used to build a document-based question-answering system. It quickly searches and connects relevant information in large documents to generate accurate answers.
Ollama + Gemma3
A generative LLM that can be executed in a local environment, providing secure AI capabilities without relying on external cloud dependencies.
E5 Embeddings
A multilingual vector processing embedding model that effectively vectorizes documents in various languages, including Korean.
RDFLib
A library for ontology reasoning, used to establish and analyze the semantic relationships between equipment, quality, and safety.
Tesseract OCR
Technology for extracting text from drawings, digitizing important information from image-based equipment drawings.
SpeechRecognition
Technology for converting speech to text, used to store and analyze voice recordings of field workers.
The DX-AI Platform combines the latest open-source technologies and in-house developed technologies to build a powerful and flexible system. Each technology is responsible for a specific function and operates organically as part of the overall system.
8. Effectiveness by Job Function
This is a practical AI agent that can be felt by all departments. By providing specialized functions for each job function, it improves work efficiency and enhances safety.
8. Scalability
Multinational Factory Application
It can support multiple languages, allowing it to be used in global manufacturing environments. It provides the same quality of service without language barriers.
Diverse Industry Domains
It can be expanded into various industry sectors such as pharmaceuticals and semiconductors. The ontology and knowledge base can be adjusted to suit the characteristics of each industry.
Higher-order Reasoning
By integrating ontology and RAG, it can perform complex causal analysis and prediction. It provides in-depth knowledge inference beyond simple search.
System Integration
It can be expanded into an integrated system by connecting with existing MES, ERP, CMMS, and others. It accelerates the digital transformation of the enterprise.
The DX-AI Platform is a flexible system that can be expanded in various directions beyond its current capabilities. It can continuously evolve and provide value in line with the growth and changes of the enterprise.
9. Summary & Proposal
Knowledge Partner, not just a chatbot
The DX-AI Platform is a specialized smart knowledge partner for manufacturing sites, not just a simple chatbot. It integrates and manages equipment, quality, and safety comprehensively.
Pilot Build
Pilot projects can be conducted on specific lines or processes to verify the effects and optimize the solution.
Collaborative Development
Customized solutions can be jointly developed and implemented to meet the customer's requirements and environment.
Investment Partnership
Through a long-term collaborative relationship, continuous technology development and system enhancement can be pursued together.
The DX-AI Platform is a core solution for the digital transformation of manufacturing sites, and it delivers optimal results through close collaboration with customers. It creates real value through a step-by-step approach and continuous improvement.
Connecting Existing Projects with the DX-AI Platform - Generating Tangible Synergies
1
Quality Prediction System
The AI agent automatically analyzes and explains the causes, conditions, and impact of the prediction results, and generates evidence by linking equipment-specific logs and past history.
  • Automated reporting
  • Minimized interpretation discrepancies among personnel
  • Utilized for management reporting
2
SEM-EDS Analysis
The AI interprets the image analysis results to generate quantitative features, history-based similar cases, and automatic annotations.
  • Reduced analysis time
  • Standardized analysis results
  • Utilized as training materials for new personnel
3
PHM / ICP Model
The AI can explain the key prediction results and answer "Which equipment elements are suspected under this condition?"
  • Strengthened credibility of predictive maintenance
  • Facilitated communication with equipment managers
We connect all of the customer's data and AI infrastructure into a single integrated intelligent agent. By integrating with existing systems, we can increase investment efficiency and lower the adoption barrier.
On-Premise AI-Powered Equipment Modernization - AI Capabilities Without Cloud Dependence
Local Execution Architecture
The DX-AI Platform operates on a fully local execution architecture, not cloud-based.
All data processing and AI inference are performed within the customer's internal systems.
Core Asset Protection
Equipment diagrams, work manuals, quality records, and regulatory data are all critical business assets.
These important information are thoroughly protected to prevent external leakage.
Closed Network Operation
Our system operates entirely within the customer's closed network - documents, queries, and responses.
It is designed to function perfectly without any internet connection.
The biggest concern for manufacturing companies is information leakage. The DX-AI Platform addresses these concerns and provides a security-first solution.
You can enjoy the benefits of AI while securely protecting your sensitive business information.
Start Small, Prove Impact, and Expand Widely
Step 1: Pilot Test (PoC)
Validate the technology and prove initial impact through a pilot AI agent test based on key drawings/documents.
Step 2: Expand to Specific Equipment/Line
Integrate OCR + IoT + Ontology Reasoning and apply to a specific equipment or production line to validate the impact.
Step 3: Integrate Across All Equipment/Quality
Expand AI capabilities across the entire equipment and quality systems by integrating with CMMS/QMS.
Step 4: Scale Across Departments/Plants
Support the enterprise-wide digital transformation, including regulatory compliance and safety management.
The key is to validate sustainable impact, not just large-scale deployment. Through a step-by-step approach, we can minimize risks and expand to the next phase based on the results of each stage. This helps improve investment efficiency and facilitate organizational adaptation.
The DX-AI Platform is an AI partner that understands the language of equipment, drawings, and regulations
Domain Expertise
We are domain experts with over 30 years of field experience. We understand how equipment condition, quality history, and regulatory standards are interconnected.
Industry Knowledge-Based Design
We are not just developers. We design AI models not by simple tuning, but by leveraging industry knowledge. This allows us to provide the features and information that are truly needed in the field.
Clear Answers
For questions that general AI cannot answer, we provide clear responses. Based on our industry-specific knowledge and ontology, we deliver accurate and practical information.
The true value of AI lies not in the technology itself, but in a deep understanding of the domain. The DX-AI Platform is a specialized partner that comprehends the language of the manufacturing field and provides the information and insights that are actually needed on-site.
We compete with "field applicability" rather than technology
We are not a platform, but a strategic partner for our customers' sites. Unlike large AI platforms, the DX-AI Platform deeply understands the characteristics and requirements of the manufacturing site and provides solutions tailored to them. We focus on the practical application and value creation in the field, rather than the technology itself.
Customized AI that understands the field, rather than large corporate solutions
The DX-AI Platform provides a field-centric, customized approach compared to large corporate solutions. It can be flexibly adjusted to the customer's specific requirements and environment, and has strengths in security and localization. It is a practical solution focused on solving real-world problems in the field.
When Cloud AI Makes You Uneasy,
the DX-AI Platform is Here
100%
Information-Centric
All data operates within the internal network only
4 Stages
Stepwise Approach
Start small, verify effects, scalable
30+ Years
Domain-Based Design
Structure and rule set based on field understanding
100%
Collaborative Partnership
Customization with customer empathy
We are not just a technology provider. We are an AI partner who upholds your operational philosophy and security principles. The DX-AI Platform will be a reliable partner that safely and effectively supports the digital transformation of your manufacturing site.