Built on an industrial-model-driven low-code platform as its agile foundation, we constructed a Smart Operations Control Tower from scratch covering three factories and 20+ systems — achieving the lean digital leap from "software for the factory" to "a factory of software"
A world-leading flexible printed circuit (FPC) manufacturer and a core Apple supplier
Customer M is a world-leading flexible printed circuit (FPC) manufacturer, dedicated to supplying Apple, Tesla and other top international technology companies with high-precision FPC products. Its products are widely used in core product lines including the iPhone, iPad, Apple Watch, and new energy vehicles.
Founded in 1984, the company was wholly acquired by its parent company in 2016 and became a subsidiary of a publicly listed company. Its Yancheng plant is the world's largest single-site FPC factory, with an annual production capacity of tens of millions of square meters of FPC.
From data analytics to the industrial data platform — seven years of continuous dedication to Customer M's digital upgrade
From an informatized factory to a smart factory — digitalization advancing in step with business scale
Aiming to "become a world-leading intelligent manufacturing enterprise," Customer M defined a clear smart factory evolution roadmap. Over the past decade, revenue, headcount, and digital talent have all grown rapidly in parallel, while information systems evolved from basic ERP/MES to data-driven real-time analytics, digital twins, and self-learning optimization.
From "software for the factory" to "a factory of software" — a four-step path to building a real-time digital world
Customer M's digital transformation vision is to build a real-time digital world covering the entire factory and supply chain. Through the "tower building" philosophy, data-driven management leaps from TQM (Total Quality Management) to total digital management — making everyone a tower builder.
Establish a multi-factory data management system; through three-factory data integration, achieve QCDSM-centered metric transparency and real-time alerting
Build root cause analysis capabilities on the big data platform; scenario-based metric models deliver drill-down insight into "why it happened"
Automatic work order triggering and execution closed loop; management and knowledge closed loops as twin drivers, achieving the full "detect → resolve → prevent" process
Integrate AI LLMs; knowledge graphs and predictive models deliver autonomous decision recommendations, evolving toward cognitive analytics and autonomous control
Core challenges and constraints manufacturing enterprises face in their digital journey
Problems on the production floor cannot be sensed in real time; key information such as equipment status, quality anomalies, and capacity bottlenecks lacks transparent presentation. Managers operate "flying blind," relying on manual reports and after-the-fact checks, with severely delayed information.
After problems are found, cross-departmental collaboration is inefficient; improvement actions lack standardized processes and tracking mechanisms. Problems recur, experience never becomes organizational capability, and outcomes depend on individual heroics rather than systematic management.
Multiple factories are scattered across locations with inconsistent management metric systems; remote command lacks data support, each factory "fights its own battle," and the group level struggles to achieve unified control and standardized operations.
Data with core customers such as Apple and upstream suppliers is not connected; supply chain delivery status and quality information cannot be obtained in real time; end-to-end visibility is missing, severely hurting supply chain collaboration efficiency and customer satisfaction.
Data scattered across isolated systems
Business functions fragmented across systems
Multiple factories in different locations
Supplier production and supply chain data not connected
Centered on the Smart Operations Control Tower, building a lean digital system covering the entire chain
The Smart Operations Control Tower Leansight built for Customer M was not simply purchasing an off-the-shelf system. Following the "tower building" philosophy, it started from an agile foundation and, through model composition of the three standard products, built bottom-up a full-stack digital system covering Equipment Control Layer → Data Platform Layer → Business Middle-Platform Layer → Data Visualization Layer. For large customers, application value is realized through two-way empowerment: downward, application requirements are decomposed into standard product models for continuous iteration; upward, application compositions form new products.
Agile Foundation + Lean Collaboration + Smart Control
Continuously capture data in real time, with full QCDSM metric transparency and automatic alerts; scenario-based root cause analysis recommends optimal solutions via simulation; automatic work order triggering and full-process closed-loop improvement management
Covering the full production chain of MOM/MES/APS/WMS/QMS and more, with scenario-based innovative apps such as PDCA/SmartOffice/SPC/equipment RPA; "Lego-style" composition, with large apps splittable into multiple sub-apps
Industrial model-driven low-code platform with visual forms and process orchestration; industrial-grade big data processing kernel with single-point concurrency of 100K/sec, PB-level data storage, and unlimited horizontal scaling
An industrial-grade data governance and integration engine combining high-concurrency data processing with a rules engine to rapidly achieve data synchronization and initial cleansing. Supports data extraction, transformation, and loading from 20+ heterogeneous systems, compresses cold data into storage (HDFS), and responds to business computing needs through layered processing.
High-concurrency data processing combined with a rules engine enables rapid cross-system data synchronization and initial cleansing, with a deployment cycle < 1 month
An industrial model-driven low-code development platform with rich built-in industry components; drag-and-drop plus full configuration rapidly builds app pages and business processes. Supports industrial object modeling and accumulation, visual customization, and source code export/import; apps can be deployed independently of the platform.
Development platform separated from apps, supporting independent app deployment and execution; integrated AI LLMs let developers build apps semantically
An agile data analytics and visualization platform for manufacturing, supporting multi-dimensional analysis, KPI system building, and cockpit displays. From real-time output to OTD lead time, from quality AUDIT to equipment OEE — operational metrics across all dimensions presented transparently.
Built-in 186 manufacturing operational insight metrics and 86 business insight metrics, rapidly building full-hierarchy dashboards from factory to team level
Integration of Digital Nebula and PDCA work order closed-loop management, achieving a management system of metric transparency → alert push → work order triggering → execution closed loop
Building on Control Tower 1.0, knowledge graph integration automates problem analysis, turning individual experience into organizational capability
By decomposing application requirements into the models/modules of the three standard products, the products are continuously iterated to better fit industry needs
Composing applications to form new products; "Lego-style" composition enables rapid innovation and meets the individual needs of different customers
Group-level unified control and site-level flexible execution, building a layered, decoupled, and coordinated digital application matrix
Serving the Group and its Electronic Circuit Business Division, focusing on unified standards, unified metrics, unified collaboration, and knowledge accumulation across factories and sites.
Serving on-site execution at each factory/site, focusing on line efficiency, quality control, equipment automation, and production execution, with rapid per-site customization.
Group-level applications ensure group-wide strategic consistency, unified standards, and knowledge accumulation; site-level applications enable flexible on-site response, rapid deployment, and continuous improvement. The two tiers are bidirectionally linked through a unified data foundation and metric models: the group sets standards, sites feed back execution; sites accumulate best practices, the group replicates and scales them.
The Leansight industrial big data platform provides PB-level data storage and analytics for Customer M's 67-line EAP system
A full-chain industrial big data processing pipeline from data integration to data applications
Connects core business systems such as MES, EAP, and WMS/ERP via API Gateway and the ETL engine, achieving real-time full-volume data collection
Models data along dimensions such as quality, metrics, production, IT operations, and equipment topics, building a layered computing system
Based on the four storage engines IDDB, LGDB, LDDB, and LDFS, it delivers real-time queries on hot data, aggregate analysis of warm data, offline archiving of cold data, and concurrent file storage
Operations monitoring with real-time metrics, online dashboards (yield/output/anomalies/OEE), ad-hoc queries and complex reports, API development and system integration, and management decision cockpits
Hot/cold data tiered management, scenario-based selection, optimal balance of performance and cost
Based on the Lucene kernel, supporting full-text search over text/PDF/Word and SQL queries. Used for low-latency query scenarios such as Panel text queries.
Columnar storage with full support for SQL-92/99 standards and TPC-H. Applied to historical data storage and medium-latency query analytics.
Based on the Hadoop big data ecosystem, applied to historical data archiving and high-latency massive data query analytics.
Suited for small file and image storage, with a universal HTTP interface. Applied to production-line test files and test image storage.
One-stop data operations, query, and management, freeing IT staff to focus on the business
Distributed agents collect server CPU/memory/IO in real time, with full-platform operations visualization
IT creates business topics and mappings; business users see recognizable names and get self-service visual queries
When data extraction tasks complete, the system notifies business users via email and messages
LeanBI supports document/parallel/distributed databases; select data sources on demand to develop dashboards
The Leansight Lean Data Platform built complete data lifecycle management for Customer M's EAP system, from real-time queries to long-term archiving. With the hot/cold tiered architecture of the four distributed storage engines, it supports storage, query, and analytics of PB-level data across 67 production lines at the optimal cost-performance ratio, while providing a solid foundation for machine learning and data mining.
Core value improvements delivered by the tower-building practice
The low-code platform builds apps in 2-4 weeks; standard products deploy in <1 month, responding agilely to business changes
100+ industrial apps out of the box, 10,000+ industrial models freely combined, and large apps splittable into sub-apps for flexible orchestration
Single-point concurrency of 100K/sec, query volume of 1 billion records, PB-level data storage, unlimited horizontal scaling and dynamic expansion
Apps can be deployed and run independently of the platform, avoiding vendor lock-in; a unified platform architecture simplifies integration complexity
Knowledge graphs structure and retain problem experience, turning individual expertise into organizational capability and reducing dependence on key personnel
Integrated AI LLMs support semantic app building; industrial scenario-based AI models enable use cases such as predictive maintenance and scheduling optimization
Industrial Model-Driven Low-Code Platform · From "Software for the Factory" to "A Factory of Software"
100+ industrial apps out of the box and 10,000+ industrial models freely combined, covering the full production chain of MOM/MES/APS/WMS/QMS and more
Continuously captures data in real time; QCDSM and man-machine-material-method-environment metrics are fully transparent with automatic alerts, shifting from after-the-fact management to proactive prevention
Scenario-based root cause analysis recommends optimal solutions via simulation, answering "why it happened" and "how to improve"
Automatic work order triggering and predictive metric alert pushes; full-process improvement action management achieving both management and knowledge closed loops
Continuous innovation in scenario-based apps such as PDCA/SmartOffice/ASN/SPC/equipment RPA control/supply chain management analytics/industrial knowledge graphs
Integrated AI LLMs let developers build apps semantically; industrial scenario-based AI models enable scenarios such as predictive maintenance and scheduling optimization
7 iron rules + 5 red lines distilled from Customer M's 8-year digital practice — the complete story from an IT manager being harshly criticized in 2017 to being promoted to factory general manager in 2024
Using tree growth as a metaphor, dynamically presenting the two-way empowerment journey of the Smart Operations Control Tower from seed to fruit