Our work is a concrete example of the application of Large Language Models (LLMs) to industrial engineering challenges. The prediction of virtual loads for reliability estimation in the automotive industry is a critical, yet highly complex and time-consuming process. Traditional methods demand significant manual effort from engineers: Selecting representative subsets of available real-world data for simulation input, developing physical models for component loads and configuring advanced simulation pipelines. This research investigates how LLMs can be leveraged as effective collaborative partners to simplify and accelerate these demanding engineering workflows. To address this, we have developed a multi-agent system, designed specifically for the industrial applications at Bosch Research. This system utilizes specialized agents to assist the engineer in critical workflow stages. These stages include: a) formulating physical equations for modeling component loads, b) fitting parameters from test drives, and c) executing the pipeline of stochastic velocity and virtual load simulations based on real-world driving data. The system is designed to handle these tasks not as a replacement, but as a powerful tool for the engineer. This presentation will discuss the current results and its effectiveness in reducing complexity and development time.