Major Swiss Bank

The bank’s innovative SafeChat project established a comprehensive GenAI governance framework, turning potential AI security challenges into a scalable, enterprise-level technological advantage.

Challenge

• Client wanted to introduce GenAI into the company with a first lighthouse use case. The priorities for customer were:
  • ensure that the employees have a compliant and secure way to interact with an LLM Chatbot
  • mitigate the risk of data leaks from employees using authorised solution like ChatGPT with private accounts
  • build the technical foundation and acceptance to build more advanced use cases in the future
• There was no established framework for secure generative AI service integration

Solution

• OpenAI Service Integration into the bank’s official Azure infrastructure
• Developed SafeChat application using internal Java application development framework including:
  • Detailed approvals
  • Robust documentation
  • Comprehensive support organization
• Extended SafeChat with RAG (Retrieval-Augmented Generation) Chatbot for two specific use cases

Business Impact

• Established first GenAI service platform acting as a security and compliance gateway
• Created a scalable model for future GenAI use cases across the bank
• Enabled secure and compliant generative AI interactions
• Successfully deployed SafeChat to over 6,000 users
• Built technical and conceptual foundation for enterprise-wide AI integration
• Mitigated risks associated with unauthorized AI tool usage