To test the potential of generative AI (GenAI), SV Informatik, an insurance IT solutions provider, wanted to optimise customer service processes as the company received many daily tariff queries.

Due to a historically evolved application, retrieving information was time-consuming and required precise keyword input to locate the desired information. SV Informatik approached Zühlke to conduct a technical proof-of-concept (PoC) and evaluate the use case's potential. The project also aimed to empower internal stakeholders with new capabilities.

Within a few weeks, we developed a technical PoC for SV Informatik, designed to assist the customer service team in swiftly and accurately responding to tariff-related queries. The GenAI application was based on an internal document repository (contract conditions, lexicons) and was extendable to include additional documents and databases. The solution featured a natural language chat interface that referenced relevant tariff documents, enhancing the ease of information retrieval.

Within just five weeks of development, there was a marked improvement in information retrieval for the customer service department. The users and the project team were happy with the solution, highlighting its effectiveness and usability.

The PoC laid the foundation for further developments with potential customer-specific expansions. The context-based (semantic) search enabled keyword-independent information retrieval, making the process easier and more flexible, and inclusive of additional user groups.