# Automated Testing of the GUI of a Real-Life Engineering Software using Large Language Models

**Authors:** Tim Rosenbach, David Heidrich, Alexander Weinert  
**arXiv:** [2505.17839](https://arxiv.org/abs/2505.17839)  
**Date:** 2025-05-23  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2505.17839v1/

## Key points

- Testing the final product with real users is crucial in software development to identify issues and inconsistencies.
- Manual testing can be time-consuming, leading to the introduction of GERALLT, a system that uses Large Language Models (LLMs) for autonomous exploratory tests on GUI.
- GERALLT generates a comprehensive list of potential unintuitive and inconsistent elements within the interface, streamlining the testing process and enhancing efficiency.
- Evaluation of GERALLT on real-world engineering software showed successful identification of interface issues, providing valuable insights for future development.
- The research paper by Tim Rosenbach, David Heidrich, and Alexander Weinert discusses how GERALLT automates GUI testing using LLMs, presenting its effectiveness at the A-Test Workshop during ICST'25.

## Summary

In the realm of software development, a crucial aspect is the testing of the final product with real users to identify any potential issues or inconsistencies in its behavior and interface. This process not only helps in uncovering unintuitive aspects of the software but also provides valuable feedback for further development. However, conducting these tests manually can be time-consuming. To address this challenge, a system called GERALLT has been introduced. GERALLT leverages Large Language Models (LLMs) to autonomously conduct exploratory tests on the Graphical User Interface (GUI) of an actual engineering software. By utilizing LLMs, GERALLT can generate a comprehensive list of potential unintuitive and inconsistent elements within the interface. The architecture of GERALLT is designed to streamline the testing process and enhance efficiency. To evaluate its effectiveness, GERALLT was applied to a real-world scenario involving an engineering software that had already undergone extensive testing by both developers and users. The results demonstrated that GERALLT successfully identified issues within the interface, providing valuable insights to support future software development endeavors. Authored by Tim Rosenbach, David Heidrich, and Alexander Weinert, the research paper titled "Automated Testing of the GUI of a Real-Life Engineering Software using Large Language Models" delves into how GERALLT revolutionizes GUI testing through automation. Presented at the A-Test Workshop during ICST'25 and published in the 2025 IEEE International Conference on Software Testing, Verification and Validation Workshops (ICSTW), this work showcases the potential of leveraging advanced technologies for enhancing software development processes.