…of employees consider the chatbot helpful for their work.*
1 min read
From Proof of Concept to Global GenAI Solution
It started with an idea - and a non-scalable prototype
An internal proof of concept was intended to show how Generative AI could help make company knowledge more readily available. The potential was recognizable - but the PoC was not ready for production.
The Challenge:
Challenges in ensuring scalability
Lack of operational stability
Insufficient data protection
Limited monitoring and lack of governance
The assignment:
An internal GPT chatbot that makes knowledge available.
We were commissioned to raise the solution to an enterprise level- technically, operationally and on the user side. An AI platform provides the technical basis.
The Solution:
A chatbot available to more than 200,000 employees
Scalable via Azure OpenAI and App Service
Multi-model support (GPT-4.1 and GPT-4.1 mini)
Privacy-compliant through automated data deletion routines
Feature enhancements such as follow-up questions, dark mode, and chat with your own documents
Access control via user groups
The AI platform - the technical foundation:
Wir entwickelten zusätzlich die Plattform, auf der der firmeninterne Chatbot läuft – und auf der weitere spezialisierte GPTs entstanden sind:
Individuelle AI-Bots für Fachabteilungen
Gemeinsames Logging, Monitoring und Alerting
Developer Experience mit Fokus auf Erweiterbarkeit
Business Impact:
%
…of employees consider the chatbot helpful for their work.*
%
…of employees feel that GPT saves them up to 2.5 hours per week.*
Net Promoter Score
%
The company’s internal GenAI chatbot is now available to more than 200,000 employees worldwide and is used extensively. A Net Promoter Score of 47 confirms strong acceptance and tangible added value. The solution delivers significant time savings and measurably higher productivity across a wide range of use cases. More than 70% of respondents believe that the chatbot improves the quality of their work results. Thanks to the AI platform strategy, the solution can also be flexibly extended—for example, to support additional specialized AI use cases.
Facing similar challenges?
We look forward to hearing from you. Please share your request using the form, and we will get back to you promptly.