At Gjensidige, Norway's largest insurer, one of the most important tasks for analysts and data scientists is the interactive visualization and simulation of data. In use cases such as risk prediction and asset management, they have established the Python-based package Streamlit as the preferred frontend for data-driven applications. In this talk, Alexandra Diem, Head of AI and MLOps, will walk you through the process of rapid LLM prototyping using OpenAI and Streamlit to create chatbots that heavily increase the efficiency of Gjensidige's internal processes. You will meet ""Eglev"", an innovative chatbot designed to empower without displacing tasks. Learn how Gjensidige turned Eglev from idea to working demo with greater than 91% accuracy in only 6 weeks. By making their lakehouse data available for non-technical resources by predicting SQL queries from natural language, Eglev allows Gjensidige's data scientists to focus on the most important tasks.