Custom solution based on 2026-06-26 Customer A Digital Exchange Meeting Minutes

Smart Operations Control Tower
Custom Solution for Customer A

Addressing President Guo's focus on OEE full-chain data integration, OPS Review integration, low-code platform usability, cautious AI deployment, and ROI — combined with Leansight's 8-year tower-building practice at Customer M (Apple supply chain lighthouse factory) — providing a phased, actionable implementation roadmap.

8+ Years
Customer M deep collaboration period
20+
Delivered industrial applications
4 Phases
Phased implementation plan
¥23M+
Customer M annualized returns
Meeting Recap

Key Meeting Topics Recap

Based on the June 26 Customer A digital exchange meeting, distilling three core topics and President Guo's key requirements

🤖

AI and Digital Practices

Both sides agreed that AI in factory scenarios requires caution, and is currently better suited for assisting data analysis, process automation, and specific-scenario machine learning. High-quality data is the cornerstone of AI deployment.

"Agent Factory" intelligent agent platform is based on PDCA cycles and the double helix model, going beyond simple Q&A
🏗️

Digital Transformation and Platform

The industrial big data platform (Fusion) has data extraction, processing, storage, and ML integration capabilities. Low-code/no-code platforms empower business users, but limitations exist in complex customization scenarios.

Briefly introduced the Smart Operations Control Tower; AI assistant can provide "CEO Headline" operational insights
🎯

Core Challenges and Direction

The biggest obstacle is breaking data silos, unifying cross-departmental metric definitions, and driving cross-departmental collaboration. President Guo's direction this year: OEE full-chain integration → OPS Review integration → Service Operations Management Center.

"We need real solutions and how we actually implemented them — what's the biggest business pain point? How to solve it? What's the ROI?"
Targeted Response

Six Pain Points — Point-by-Point Response

For each core question raised by Customer A during the meeting, providing Leansight's targeted solutions and Customer M practice evidence

01

Data Silos and Inconsistent Metric Definitions

Customer A's current state: Data is scattered across ERP/MES/EAP/equipment systems by department, cross-departmental operational metric definitions are inconsistent, making horizontal benchmarking difficult.
"The biggest obstacle to current digital transformation is breaking down data silos between departments and unifying cross-departmental operational metric definitions"
Leansight Solution

LeanFusion Unified Data Foundation + QCDSM Metric System

  • LeanFusion Data Capability Platform: From ingestion, processing to service and runtime, making a data requirement faster to become a reliable, reusable, sustainably managed result. One pipeline spans the entire data lifecycle, building a "data capability network" — data is not "transported assets" but "orchestrable, consumable capabilities"
  • 200+ data source pre-built connectors: Covering EAP/MES/ERP/WMS/equipment IoT/SCADA/PLC and other full systems, completing heterogeneous data source integration at hourly level, unified data lake/warehouse ingestion
  • QCDSM five-dimensional metric system: Quality (Yield/Cpk), Cost (Unit Manufacturing Cost), Delivery (OTD/Cycle Time), Safety (Safety Incident Rate), Morale (UPPH/Per-Capita Output), drilling down from group level to production line level, unified definitions
  • Metric dictionary management: Every metric's definition formula, data source, calculation logic, and refresh frequency are all configuration-managed, eliminating the problem of "different departments calculating different numbers for the same metric"
  • Cross-departmental benchmarking dashboards: Same metric compared horizontally across multiple factories/lines/shifts, anomalies auto-highlighted
⚡ LeanFusion · AI Upgrade Capabilities (Intelligent Evolution of the Data Capability Platform)
  • AI-Assisted Data Modeling: Natural language generates ETL, zero-code cross-source data fusion tasks, business users can self-configure
  • Intelligent Metadata Discovery: Auto-identifies field semantics, lineage relationships, data quality scoring, making data assets "understandable"
  • Metric Semanticization and Q&A: Metrics are no longer "static numbers" — AI can perform natural language Q&A and root-cause drill-down based on the metric dictionary
  • Data Quality Self-Healing: AI identifies anomalous drifting data, auto-alerts and recommends repair strategies, moving from "post-event cleansing" to "in-process self-healing"
  • Real-Time Feature Computation: Provides millisecond-level real-time features for downstream AI models (yield prediction/fault diagnosis), no need for manual offline computation
  • Agent-Oriented API Output: All data capabilities exposed as standardized APIs/MCP, allowing CEO Headline, Agent Factory, and other upper-layer agents to directly "call data"
Customer M Practice: Customer M originally had 4 factories each maintaining Excel reports, with the same "yield" metric defined differently. After Leansight deployed LeanFusion, it unified 67 production line data sources, established a group-level metric dictionary, and monthly business meetings shifted from "reconciling data" to "analyzing data," improving meeting efficiency by 60%. Since 2022, LeanFusion has progressively integrated AI capabilities, with metadata auto-discovery coverage increasing from 15% to 85%, and average data anomaly identification time reduced from 2 hours to 8 minutes.
02

OEE Full-Chain Data Integration

President Guo's core direction this year: Integrate the full-chain data for equipment utilization rate (OEE). Currently, equipment data is scattered, unable to form OEE drill-down from equipment level → line level → factory level.
"Integrate the full-chain data for equipment utilization rate (OEE) to achieve integration of operations management (OPS Review)"
Leansight Solution

EAP Data Collection + Equipment OEE Model + Full-Chain Drill-Down

  • EAP real-time data collection: Direct connection to equipment automation platforms, second-level collection of equipment status (Run/Idle/Down/Engineering), alarm codes, output, and process parameters, covering Wire Bonding/Die Bonding/Molding/Test Handler/AOI and other packaging & testing core equipment
  • OEE three-layer decomposition model: Availability (Equipment Availability) × Performance (Performance Rate) × Quality (Quality Rate), each layer automatically attributes to specific loss reasons (material waiting/changeover/equipment failure/process tuning/quality anomaly)
  • Full-chain drill-down: Group OEE → Factory OEE → Workshop OEE → Line OEE → Equipment OEE → Shift OEE, click to drill down at any level to locate bottleneck equipment and loss causes
  • Loss tree analysis: Automatically generates OEE loss distribution Pareto charts, Top 5 loss causes at a glance, supporting management decisions
Customer M Practice: After deploying EAP data collection across 67 production lines at Customer M, OEE shifted from "monthly manual statistics" to "real-time automatic calculation," with data latency reduced from 7 days to <5 minutes. Through loss tree analysis, changeover time was found to account for 23% of OEE losses; after targeted optimization, changeover time was reduced by 35%, and overall OEE improved by 4.2 percentage points.
03

Business Departments Can't Use the Low-Code Platform

Customer A feedback: The current low-code platform is not satisfactory, and business departments have difficulty using it. They need a Leansight demo environment and product sharing.
"Low-code platform sharing — the current one customers are not satisfied with, business departments have difficulty using it"
Leansight Solution

LeanCodee Industrial-Grade Low-Code Platform — Business-Usable, IT-Controllable

  • Industrial scenario pre-built templates: 20+ industrial application templates built-in (eDoc forms/SPC dashboards/OEE cockpits/work order management/equipment inspection), business departments configure by drag-and-drop based on templates, no need to build from scratch
  • Dual-layer development mode: Business users use visual drag-and-drop for 80% of standard scenarios, IT developers use code to extend 20% of complex logic on the same platform, solving the "low-code can't do complex customization" pain point
  • Seamless integration with data foundation: LeanCodee directly calls LeanFusion's data models and LeanBI's dashboard components, business users don't need to worry about data source integration when developing applications
  • Mobile responsive: Develop once, auto-adapts to PC/tablet/phone, shop floor uses Pad for scanning, management uses phone for dashboards
Customer M Practice: Customer M developed 20+ applications through LeanCodee, of which eDoc (paperless forms) was independently built by the business department IT within 1 month, and Equipment RPA was developed and launched by equipment engineers in <3 months. The "can't use" problem for business departments was solved through a "template starting point + IT enablement" model, with 80% of applications led by business departments.
04

OPS Review Integration and Operations Management Empowerment

President Guo needs to achieve integration of operations management OPS Review, ultimately serving the operations management center to enable effective measurement of input-output.
"Ultimately serving the operations management center to enable effective measurement of input-output"
Leansight Solution

Smart Operations Control Tower — Multi-Role Insights + Meeting-Driven Closed-Loop

  • Five-role drill-down dashboards: General Manager (factory-wide operations overview) → Business Unit Director (production line performance benchmarking) → Workshop Manager (real-time production monitoring) → Team Leader (work order execution details) → Warehouse & Logistics (material readiness rate), each role sees different granularity of the same data set
  • OPS Review meeting mode: Control tower big screen directly drives daily/weekly/monthly operations meetings, auto-generates agenda data packages before meetings, real-time drill-down to locate issues during meetings, auto-dispatches Action Items and tracks closed-loop after meetings
  • Input-Output ROI dashboard: Automatically links equipment investment/labor input/material costs with output (good product quantity/output value), calculating per-line ROI, per-equipment ROI, per-person ROI, providing quantitative decision basis for the operations management center
  • "CEO Headline" operations summary: AI auto-generates daily operations headlines, understand key metric changes, Top 3 anomalies, and pending items in 5 seconds
📊 OPS Five-Layer Management Drill-Down System
Top-down metric decomposition establishes an intelligent decision system; bottom-up data贯通 breaks data silos; role-centric full participation in root cause analysis and continuous optimization; domain-perspective metric system achieves automation and transparency
Customer Visits
Government Leaders
External Display & Promotion Layer
Smart Factory Showroom Real-Time Operations Big Screen Corporate Image Display
External Window
Brand & Credibility
President
General Manager
Business Decision Layer
Performance Management President's Cockpit CEO Headline Input-Output ROI
Strategic Direction
Annual/Quarterly Decisions
Director
Business Domain Layer
Q Domain (Quality) C Domain (Cost) D Domain (Delivery) S Domain (Safety) M Domain (Morale)
Cross-Department Collaboration
Weekly/Monthly Review
Workshop Leader
Department Leader
Functional Domain Layer
Inner Layer Copper Deposition Plating Drilling Routing/Exposure AOI ET Testing
On-Site Execution
Daily OPS Review
Manager
Team Leader
Fine-Grained Domain · Refined Management
Plan Achievement Rate FTT Safety Hazards Scrap Material Usage Improvement Proposals Utilization Rate Anomaly Downtime Real-Time Output Scrap Rate
Fine-Grained Execution
Real-Time Monitoring & Response
Top-Down:Strategic goals → Metric decomposition → Task assignment → Accountability to individuals, establishing an intelligent decision system
Bottom-Up:Equipment data → Line data → Factory data → Group data, breaking data silos
Role-Centric:Five-level roles each see what they need, full participation in root cause analysis and continuous optimization
Domain Perspective:Q/C/D/S/M five-domain metric system automated and transparent
Customer M Practice: After deploying the control tower at Customer M, daily morning meetings shifted from "each department reporting for 30 minutes" to "big screen review for 5 minutes + discussion for 10 minutes," and monthly business meeting pre-data preparation time was reduced from 2 days to 0.5 days. The CEO Headline feature lets management grasp factory-wide operations status via phone every morning. The five-layer drill-down system lets the chairman drill down to any shift's specific equipment operating parameters with one click, compressing the decision chain from "weekly" to "minute-level."
05

AI Deployment Requires Caution — Data Governance Is Prerequisite

Consensus: AI in factory scenarios requires caution, with instability and "hallucination" risks. High-quality data is the cornerstone of AI deployment. Need product demos for Agent Factory, CEO Headline, and AI assistant.
"AI model training and effective application heavily depend on structured, high-quality data"
Leansight Solution

Phased AI Deployment Path — From Data Governance to Intelligent Agents

  • Phase 1: Data governance first: Through LeanFusion, complete data cleansing, standardization, and quality validation, establishing an industrial data asset catalog, providing "clean fuel" for AI
  • Phase 2: Assisted analysis AI: Embed machine learning models in LeanBI to achieve yield anomaly root-cause analysis, equipment failure prediction, and SPC intelligent early warning — these are "low-risk, high-value" AI scenarios
  • Phase 3: AI vision applications: Camera-based operational compliance detection (safety helmet wearing/equipment switch compliance/electronic fences/MOB trajectory tracking), a proven mature scenario at Customer M
  • Phase 4: Agent Factory intelligent agents: After the data foundation is solid, introduce PDCA double-helix agents to achieve self-driven anomaly discovery → analysis → decision → execution closed-loop, deeply integrated with the control tower
Customer M Practice: Customer M's AI applications strictly follow the "data first" principle, focusing on data governance and platform building from 2017-2021, only gradually introducing AI Monitor (vision safety detection) and yield attribution models after 2022. AI is not "all at once" but "naturally arrives when conditions are ripe."
06

ROI and Deliverables

President Guo's core concern: What is ultimately delivered to the customer? What real value does it bring? How much improvement, how much labor saved? What's the ROI?
"What's actually delivered to the customer? How much improvement, or how much labor saved? ROI"
Leansight Solution

Quantified Delivery + Measurable Business Value

  • Clear deliverables: Not "selling a product" but delivering "a running system + a set of dashboards + a metric system + a continuously iterating platform" — product + implementation + knowledge transfer
  • Quantified benefit tracking: Each application defines a baseline at launch, tracks key metric improvement monthly, quarterly review confirms ROI
  • Customer M proven returns: Annualized returns of ¥23M+, including 3x anomaly response efficiency improvement, 20% quality cost reduction, 15% OTD improvement, 2,000+ person-days/year saved through report automation
  • ROI: Customer M's 8-year cumulative investment of approximately ¥5M (software + implementation), cumulative directly quantifiable returns >¥180M, ROI >36x
Customer M Practice: "What we ultimately deliver is not PPT, it's systems used every day. eDoc eliminated paper from the shop floor, the control tower means management no longer asks 'where's the data?', AI Monitor means safety officers no longer stare at monitors full-time. Every application has clear Before/After comparison data."
Solution Architecture

Smart Operations Control Tower Four-Layer Architecture

Bottom-up platform building, top-down data pull — the dual-direction empowerment tower-building philosophy

🧠 Autonomous Discovery Layer — CEO Headline · Business Intelligence Engine Learn Core Concept →

For decision-makers: AI automatically scans enterprise internal and external changes daily, distilling them into insightful, evidence-based business headlines that aid decision-making. Not one more report, but missing fewer important events.

Business Signal Discovery Cross-System Correlation Analysis Fact/Inference/Pending Classification Multi-Agent Architecture Enterprise Profile Learning Business Meeting Direct-Drive
▲▼ Bidirectional Data Flow

🎯 Smart Control Layer — Business Decision Brain

For management: QCDSM cockpit, multi-role drill-down dashboards, input-output ROI dashboard, OPS Review meeting-driven

QCDSM Cockpit OPS Review ROI Dashboard Cross-Factory Benchmarking Multi-Role Drill-Down Agent Factory (Phase 4)
▲▼ Bidirectional Data Flow

⚙️ Lean Collaboration Layer — Operations Execution Hub

For business departments: OEE monitoring, work order closed-loop, PDCA problem management, SPC process control, low-code application matrix

OEE Full-Chain Work Order Closed-Loop PDCA SPC eDoc Forms EMS Equipment Management AI Monitor Low-Code Applications
▲▼ Bidirectional Data Flow

🏗️ Agile Foundation Layer — Data and Development Platform

For IT and data teams: unified data fusion, low-code development engine, BI analytics components, EAP big data storage

LeanFusion Data Fusion LeanCodee Low-Code Engine LeanBI Analytics Components EAP Big Data Platform Metric Dictionary Management Data Quality Validation
Beyond the Ask

Beyond the Question: Leansight Core Advantages

Not only solving Customer A's current pain points, but bringing industrial-grade platform capabilities and a long-term evolution path

🧩

Tower-Building Philosophy: Dual-Direction Empowerment

Top-down using data to pull business requirements, bottom-up using platform building to quickly meet requirements. Not "do a project and leave," but "build a tower that can continuously grow."

Continuous Evolution Lego-Style Composition
📊

EAP Massive Data Processing Capability

67 production lines, 221TB annual peak database, 120GB/month graphic files. Four-layer hot-cold tiered storage (Ultra-Hot IDDB/Hot LGDB/Warm LDDB/Cold LDFS), 200K TPS, PB-level storage, 1-billion-record query <1 second.

200K TPS PB-Level Storage <1s Query
⚡

LeanFusion AI Upgrade: Intelligent Evolution of the Data Capability Network

Upgrading from "data fusion platform" to "data capability platform" — AI-assisted modeling, intelligent metadata discovery, metric Q&A, data quality self-healing, agent-oriented API output. Providing "callable data capabilities" rather than "transported data assets" for CEO Headline and Agent Factory.

AI Modeling Data Q&A Agent Ready
🏭

Industrial-Grade Low-Code: Business-Usable

Unlike general-purpose low-code platforms, LeanCodee has 20+ built-in industrial application templates. Business departments start from templates, IT extends complex logic, 80% of applications are led by business departments. Solving the "business can't use it" pain point.

20+ Templates Dual-Layer Dev Mobile Responsive
🔬

Phased AI Deployment: Data First

Not blindly pushing AI, but a four-step approach: "data governance → assisted analysis → vision applications → intelligent agents." Each step has Customer M validation cases, each step has quantifiable returns, eliminating "AI hallucination" risk.

Data First Low Risk High Value Agent Factory
🌐

Group & Site Dual-Layer Application System

Group-level applications with unified standards (QCDSM/PDCA/SmartOffice), factory-level applications with flexible customization (eDoc/SPC/OEE). Unified data foundation with bidirectional linkage, ensuring both management consistency and factory autonomy.

Group Unified Factory Flexible
⏱️

Rapid Launch: Weekly Delivery

Standard products + templated implementation, simple applications launch in 1 week, medium complexity <3 months, group-level systems 3-6 months. Not "wait a year to see results," but "quick wins, continuous iteration."

1-Week Launch Fast Iteration
Benchmark Case

Customer M Case: Complete Closed-Loop from Pain Points to Value

Apple supply chain lighthouse factory, 8-year deep collaboration, answering President Guo's three core questions

❓ President Guo's Three Questions

  • What's the biggest pain point from a business perspective? How to solve it?
  • What problems did we encounter during the project, and how did we solve them?
  • How to drive a department or even the entire company to do this?
  • What's ultimately delivered to the customer? What real value does it bring?

💡 Customer M's Real Answers

  • Biggest pain point: 4 factories with siloed data, monthly business meetings "reconciling data" rather than "analyzing data" → Deployed LeanFusion unified data foundation
  • Project challenge: Business departments thought "IT systems are IT's business," reluctant to cooperate → Started with eDoc (paperless forms), letting front-line workers immediately feel the value of "less paperwork," building trust before pushing complex applications
  • Driving strategy: First used control tower big screen to show top management the value (top-down pull), then used low-code platform for business departments to self-develop (bottom-up build), forming a positive cycle of "top management wants it, business uses it, IT supports it"
  • Deliverables: 20+ running application systems, a unified metric system, a continuously iterating low-code platform, annualized returns of ¥23M+
2017

Data Analytics Kickoff

Started with QCDSM core metric dashboards, establishing the first cross-system data visualization. Let management "see" unified operational data for the first time.

Delivered: 5 core metric dashboards | Pain point: Manual data aggregation
2018

Smart Operations Control Tower Launched

Implemented factory-wide data-driven operations system — the Smart Operations Control Tower. Covering 4 factories and 67 production lines, achieving real-time operations monitoring and multi-role drill-down.

Delivered: Control Tower system | Pain point: Operational data not visible
2019-2022

LeanFusion Introduction and EAP Big Data Governance

Introduced LeanFusion data fusion platform, conducted EAP and industrial big data governance. Established four-layer hot-cold tiered storage, solving 221TB annual incremental data storage and query performance issues.

Delivered: EAP Big Data Platform | Pain point: Massive data couldn't be stored or queried
2022-Present

Industrial Data Platform Application Matrix

On the solid data foundation, developed 20+ production and quality-related applications through LeanCodee low-code platform: eDoc/SPC/OEE/PDCA/AI Monitor/EMS, etc., 80% led by business departments.

Delivered: 20+ application systems | Pain point: Slow business requirement response
3x
Anomaly Response Efficiency
Average time from discovery to handling reduced 66%
-20%
Quality Cost
SPC early warning + yield attribution reduced defect rate
+15%
OTD Improvement
Material readiness monitoring + delivery prediction
2000+
Person-Days/Year Saved
Report automation + RPA data handling
Phased Roadmap

Phased Implementation Plan

Four phases in sequence, each with clear deliverables and measurable value — quick wins, continuous growth

PHASE 1

Data Foundation + OEE Integration

Solve the "can't see" problem, let data flow
  • Deploy LeanFusion, connect EAP/MES/ERP core data sources
  • Establish OEE three-layer decomposition model (Availability×Performance×Quality)
  • Build equipment-level → line-level → factory-level OEE drill-down dashboards
  • Establish QCDSM metric dictionary, unify definitions
  • OEE loss tree auto-analysis
Deliverables:OEE real-time dashboards + metric dictionary + data foundation
Value:OEE data latency from days to minutes, loss causes visualized
PHASE 2

Operations Control Tower + OPS Review

Solve the "can't manage" problem, make operations meetings data-driven
  • Deploy Smart Operations Control Tower, covering five-role drill-down dashboards
  • Establish OPS Review meeting-driven mode (daily/weekly/monthly)
  • Launch input-output ROI dashboard
  • CEO Headline daily operations summary
  • Cross-factory/line/shift horizontal benchmarking
Deliverables:Control Tower system + OPS Review process + ROI dashboard
Value:Meeting efficiency up 60%, management decisions data-backed
PHASE 3

Low-Code Application Matrix + Closed-Loop Management

Solve the "can't do well" problem, let business departments self-develop
  • Deploy LeanCodee low-code platform, train business departments
  • Start with eDoc paperless forms (quick wins)
  • Progressively launch SPC/PDCA/EMS/WMS and other applications
  • Establish work order closed-loop and knowledge graph
  • EAP big data platform four-layer tiered storage
Deliverables:Low-code platform + 10+ applications + work order closed-loop system
Value:Anomaly response 3x, report automation saves 2000+ person-days/year
PHASE 4

AI Intelligent Agents + Continuous Evolution

Solve the "hard to collaborate" problem, from assisted analysis to self-driven
  • AI yield anomaly root-cause analysis (low-risk ML scenarios)
  • AI Monitor vision safety detection (operational compliance)
  • Equipment failure predictive maintenance models
  • Agent Factory intelligent agent pilot (PDCA double helix)
  • Knowledge graph-driven intelligent decision support
Deliverables:AI application suite + Agent Factory pilot
Value:Quality cost -20%, AI-assisted decisions from "seeing" to "foreseeing"
ROI Analysis

Return on Investment Analysis

Based on Customer M's 8-year practice data, estimating Customer A project ROI

Phase Investment Estimate Deliverables Expected Annualized Returns Payback Period
Phase 1 ¥500K-800K OEE dashboards + data foundation + metric dictionary OEE improvement of 2-4 percentage points, capacity release value approximately ¥2-4M/year 3-6 months
Phase 2 ¥600K-1M Control Tower + OPS Review + ROI dashboard Meeting efficiency up 60%, management decision cycle shortened 50% 6-9 months
Phase 3 ¥800K-1.5M Low-code platform + 10+ applications + work order closed-loop Anomaly response 3x, report automation 2000+ person-days/year, quality cost -15% 9-12 months
Phase 4 ¥600K-1.2M AI applications + Agent Factory pilot Quality cost -20%, equipment unplanned downtime -30% 12-18 months
Total ¥2.5-4.5M Complete Smart Control Tower System Estimated annualized returns ¥8-15M+ Overall ROI 12-18 months

* The above data is estimated based on Customer M (Apple supply chain lighthouse factory) 8-year practice benchmarks; actual ROI depends on Customer A's specific scenario scale

Deliverables

What We Deliver

Not PPT, not proof of concept — systems used every day, quantifiable value, a continuously growing platform

📦

Running Systems

Each phase delivers live systems that business departments use daily, not demos. Systems pass acceptance criteria: designated user daily active rate >80%, designated feature usage rate >90%.

Acceptable
📋

Metric System and Methodology

Deliver a complete QCDSM metric dictionary, OEE calculation model, and OPS Review meeting process. These "soft assets" belong to Customer A and can run independently even without the Leansight platform in the future.

Independent
🔧

Low-Code Platform and Training

Deliver LeanCodee low-code platform usage rights and training, enabling Customer A business departments to self-develop new applications. Training covers template usage, data integration, and dashboard building, ensuring "teaching to fish."

Self-Sufficient