Live Production System

This system is already running.
For a 200,000+ visitor venue.

Everything shown in the AI Assistant and Smart Inbox is built on the same architecture that powers a live analytics infrastructure for a major Dutch public facility. Built and maintained by one person.

All numbers on this page are fetched live from the production database
--
Revenue Tracked
live from POS system
--
Visitors Tracked
hourly resolution
--
Database Tables
PostgreSQL
--
Live Records
10+ data sources
Live System Status

Data Sources (10 connected)

Point-of-Sale (ReCreateX)LIVE
Staff Scheduling (Teamplanner)LIVE
Building Sensors (Arrigo, 19)LIVE
Energy + Solar (Van Dorp)LIVE
Google Reviews + AnalyticsLIVE
Shower Sensors (Acqua-Net)LIVE

AI-Powered Systems (10 active)

AI Morning BriefingClaude, daily 06:30
Visitor Forecast (V4)7 segment models
Swim Lesson Churn Detection619 students
Review Auto-Response24-point system
Staff Optimization Enginedemand-based
Internal AI Chatbot (MCP)24 tools
Integration Ecosystem
R
ReCreateX
POS + Ticketing
T
Teamplanner
Staff Rostering
A
Arrigo
Building BMS
V
Van Dorp
Energy + Solar
G
Google
Reviews + GA4
A
Acqua-Net
Shower Sensors
S
SolarEdge
229 Devices
N
NexusPortal
Swim Lessons
O
Open-Meteo
Weather Forecast
n
n8n
Workflow Engine
AI Systems in Production

Morning Briefing

Every day at 06:30, AI analyzes yesterday's performance, today's forecast, weather, staff gaps, and open issues. Delivered as a concise briefing to management.

Claude Sonnet // n8n pipeline // push notification

Visitor Forecast V4

NegBin regression model with 7 segments. Predicts visitors per hour, accounting for weather, holidays, events, and historical patterns. R2 = 0.427.

Python // scikit-learn // 442 days training data

Quality Gate Engine

9 automated checks that validate data freshness, forecast drift, reconciliation, and anomalies before any number reaches the dashboard.

9 gates // auto-blocking // confidence scores

Churn Detection

Monitors 619 swim lesson students for attendance patterns, progress stalls, and engagement signals. Flags at-risk students before they leave.

Subscription analytics // early warning // NexusPortal

Review Auto-Response

24-point framework that classifies Google reviews, generates proportional responses, and escalates issues. Never defensive, always on-brand.

Claude // 24-point framework // tone calibration

MCP AI Chatbot

Internal chatbot with 24 specialized tools. Staff asks questions in natural language, gets answers from live data. Revenue, visitors, staff, energy.

MCP protocol // 24 tools // Claude Sonnet
Build Timeline
SEP 2025 - JAN 2026
Foundation
Database architecture, POS integration, visitor tracking, energy monitoring. First dashboard with real-time KPIs.
FEB 2026
Intelligence Layer
Forecast model V4 with 7 segments, quality gate engine, AI morning briefings, swim lesson analytics.
MAR 2026
Automation
Decision engine, staff optimization, review auto-response, MCP chatbot with 24 tools, push notifications.
APR 2026
Referee Abroad AI Platform
Same architecture adapted for tournament operations. AI Assistant, Smart Inbox, knowledge base. Built in days, not months.
Technology Stack
P
Python
Backend
F
Flask
API Layer
P
PostgreSQL
Database
C
Claude AI
Intelligence
n
n8n
Workflows
D
Docker
Infrastructure
What This Means for Referee Abroad

Same architecture. Adapted for tournaments.

The AI Assistant and Smart Inbox you tried are production-ready implementations of the same technology. Here is the direct translation:

AI Assistant = same Claude engine
Smart Inbox = same classification AI
Morning Briefings = same daily pipeline
n8n Workflows = same automation hub
Command Center = same dashboard
Federation Reports = same PDF export

Ready to see what this can do for Referee Abroad?

The demos show working examples on real infrastructure, not mockups. And this is just the beginning — the same architecture supports tournament briefings, automated workflows, sponsor reporting, and predictive analytics.