Back office ops · Production

METZ CONNECT deploys DeutschlandGPT for secure enterprise AI, achieving 85% time savings on data warehouse migration

The problem

METZ CONNECT employees wanted to use AI for daily work, but strict internal data security policies clashed with uncontrolled external AI usage. Initial attempts at a compromise using Microsoft Copilot and ChatGPT quickly led to a loss of control over who was using which AI tools and how, raising concerns that sensitive company data could reach insecure external systems.

First attempt

A partial release of Microsoft Copilot and IT's use of ChatGPT as a middle-ground solution quickly resulted in a loss of oversight, leaving it unclear who was using which AI tools, and prompting IT to block all external AI services.

Workflow diagram · grounded in source
1
AI demand vs. security policy conflict
trigger
“Der Wunsch der Mitarbeiter, KI-Technologien für ihre tägliche Arbeit zu nutzen, stieß auf die strengen internen Datensicherheitsrichtlinien des Unternehmens”
2
Confluence and Jira integration
integration
“Integration der Systeme Confluence und Jira über vorhandene Standardschnittstellen”
3
AI automates database migration
ai_action
“Manuell hätte dieser Prozess etwa drei Wochen Arbeitsaufwand für drei Personen bedeutet. Mithilfe von KI unterstützten Spezialanwendungen konnten jedoch die neuen Tabellen automatisiert mit passenden Spaltennamen angelegt und die Replika…”
4
Key-user multiplier training
feedback_loop
“Da nicht jeder Mitarbeiter tiefgehende KI-Kenntnisse besitzt, konnten diese Multiplikatoren maßgeschneiderte Lösungen für ihre Abteilungen entwickeln und das Wissen intern weitergeben. Dieser dezentrale Ansatz entlastete die IT-Abteilung…”
5
Special applications deployed
output
“Spezialanwendungen in Entwicklung, die Texte fachgerecht, vor allem in Fremdsprachen, übersetzen sollen. Zudem wurden bereits mehrere Spezialanwendungen entwickelt, die im täglichen Arbeitsalltag Zeit einsparen, wie beispielsweise Tools …”
Reported outcome

IT regained full control over AI usage with GDPR compliance ensured and sensitive company data protected.
The data warehouse migration was completed in one week instead of three weeks for three people, representing 85% time savings. Additional special applications were deployed for translation, meeting minutes, email revision, and helpdesk ticket creation.

Reported metrics
Data warehouse migration time savings85 %
Data warehouse migration durationVon 3 Wochen (3 Personen) auf 1 Woche reduziert
Reported stack
DeutschlandGPTMicrosoft CopilotChatGPTConfluenceJira
Source
https://www.deutschlandgpt.de/case-studies/metz-connect
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

IT regained full control over AI usage with GDPR compliance ensured and sensitive company data protected.

What tools did this team use?

DeutschlandGPT, Microsoft Copilot, ChatGPT, Confluence, Jira.

What results were reported?

Data warehouse migration time savings: 85 %; Data warehouse migration duration: Von 3 Wochen (3 Personen) auf 1 Woche reduziert (source-reported, not independently verified).

What failed first in this deployment?

A partial release of Microsoft Copilot and IT's use of ChatGPT as a middle-ground solution quickly resulted in a loss of oversight, leaving it unclear who was using which AI tools, and prompting IT to block all extern…

How is this back office ops AI workflow structured?

AI demand vs. security policy conflict → Confluence and Jira integration → AI automates database migration → Key-user multiplier training → Special applications deployed.

Related back office ops cases