EMSTEEL · Digital Operations Command Centre

⬡

Why This POC Exists

The EMSTEEL Digital Operations Command Centre demonstrates how factory operations, Microsoft cloud services, Azure FinOps, Security & Identity, and AI-governed recovery can be unified into a single operational decision platform. The objective is to eliminate siloed dashboards and accelerate decision-making through correlated operational intelligence — so that the right person sees the right signal and can act in seconds, not hours.

Problem Solved

Siloed dashboards force operators to cross-reference 5+ tools to understand a single incident.

Value Delivered

One correlated view: incident → root cause → financial impact → AI recovery → approval.

Audience

EMSTEEL executives, operations teams, platform engineers, and security leads.

Core Scenario — Plant 2 Incident

All pages tell one connected operational story. This is the thread.

Plant 2 production degradation→Azure telemetry bottleneck→AED 2.1M production risk→Cloud cost anomaly→Identity/security exposure→AI-governed recovery

Incident ID

INC-EMST-2026-0625-PLANT2

Location

Abu Dhabi Industrial City

Affected Area

Plant 2 / Hot Strip Mill / Rolling Line 2

Monthly Exposure

AED 2,100,000

Shift Impact

AED 185,000

Output Loss

142 tonnes / day

Recovery Estimate

12 minutes

AI Confidence

97% root cause

Correlated Signals

10 signals

Impacted Systems

8 systems

Root Cause

Duplicate telemetry stream

Resolution

RefleX autonomous recovery

Persona Guide

Personas change the operational lens, not the underlying facts. All four personas see the same Plant 2 truth — through a different decision context. Switch personas using the topbar selector.

◈ExecBusiness risk and decision-making lens

Financial exposure (AED)
Recovery confidence
Decision required now
Outcome of action vs inaction
Board-level governance

⚙OpsFactory operations and production lens

Plant health and throughput
Shift readiness
Recovery sequence and ownership
Production continuity
Rolling Line 2 status

◇PlatformAzure cloud and reliability lens

IoT Hub capacity and telemetry flow
Service dependencies
Pipeline recovery steps
Reliability thresholds
Capacity economics

⬡SecurityIdentity risk and control lens

Privileged access exposure
Production-system risk surface
Security action trail
Containment timeline
Auditability

Page Guide

⊞

Mission Control →

Executive operational overview — KPIs, situation summary, recovery actions

Recommended use cases

Executive risk review
Decision-making under pressure
Recovery prioritisation

△

Incident Intelligence →

Root cause analysis, AI investigation narrative, and impact timeline

Recommended use cases

Incident review and RCA
Major incident briefings
Board-level impact summary

◈

Cloud FinOps →

Azure cost intelligence linked directly to operational events

Recommended use cases

Cost anomaly analysis
Savings identification
Business-cost alignment

⬡

Security & Identity →

Identity risk correlation across the Plant 2 incident

Recommended use cases

Access reviews
Security investigations
Risk containment prioritisation

⚡

RefleX Operations →

AI-assisted recovery execution with human approval workflow

Recommended use cases

Automated remediation
Human approval decision
Operational execution audit

◉

Estate Operations →

Estate-wide plant and system dependency map

Recommended use cases

Impact analysis
Plant-to-system relationships
Cascading risk review

◎

AI Operations →

AI governance controls, autonomy settings, and full audit trail

Recommended use cases

AI authority governance
Operational control review
Auditability and compliance

⊙

Help & POC Guide →

Platform understanding, onboarding, and demo preparation

Recommended use cases

Stakeholder handover
Demo preparation
Developer onboarding

Simulated vs Production-Ready

Simulated in POC

◎Seeded Plant 2 incident and telemetry
◎Seeded Azure cost anomaly data
◎Seeded identity and security events
◎Demo-mode RefleX approval workflow
◎In-memory AI audit log
◎Mock Lumi intelligence (backend API)
◎Static estate topology
◎Synthetic production forecasts

Production Integration Candidates

✓Azure Monitor + IoT Hub telemetry
✓Microsoft Sentinel (SIEM)
✓Microsoft Entra ID (identity)
✓Defender for Cloud
✓Azure Cost Management API
✓ServiceNow / Jira ticketing
✓SAP and MES production data
✓SCADA historians and OPC-UA
✓Azure Event Hub event streaming
✓CMDB and asset inventories
✓Real approval workflow engines
✓Audit persistence and compliance

Developer Notes

Architecture

Next.js 14 App Router frontend (TypeScript + TailwindCSS) + FastAPI backend (Python). Fully containerised via Docker Compose. No external cloud dependencies in demo mode.

Seeded Data Approach

All incident, telemetry, cost, and security data is seeded in backend/app/data/*.json. The Plant 2 scenario is the canonical reference. Data is deterministic and consistent across all pages.

Persona System

Personas (Exec, Ops, Platform, Security) are client-side context stored in localStorage. The PersonaLens component renders page-specific framing for each persona. All four see the same underlying data — different emphasis only.

Lumi AI Intelligence

Lumi is backed by /api/copilot on the FastAPI backend. Responses are structured (summary, root_cause, business_impact, recommended_actions, confidence). In POC mode, responses are seeded but structurally identical to a production LLM response.

RefleX Operations

RefleX operation state is managed in-memory (GovernanceContext). Approval actions trigger state transitions but do not write to disk. The pattern is compatible with a real workflow engine (ServiceNow, Jira, etc.) via backend API replacement.

Enterprise Roadmap

For production: add Microsoft Entra ID authentication, RBAC by persona, tenant isolation, real connector integrations, persistent audit logs, observability (OpenTelemetry), and production configuration management.