Claus Schopf
Senior Test Automation Engineer and Software Developer at T-Digital
BIO
Claus Schopf is a 57-year-old physicist living in Munich, married and has two daughters and one cat.
He’s working in Deutsche Telekom since 1998 where he had many different roles in software testing, software development, process design and management. Most recently he is working in AI projects being Member of the AI4Testing Initiative. Also he is giving AI training for the internal DevOps Academy.
Talk (Advanced):
AI-powered migration of a complex commercial test automation framework to Java Playwright
To implement tool standardization in our company the focus is on the migration of legacy solutions (with license costs) to state-of-the-art (open source) solutions. In this case we work on the migration of UFT test automation frameworks to Java Playwright with the help of AI. This is particularly difficult because the program structures of the source and destination are quite different. These commercial frameworks include more than 100 automated test case chains, each spanning around 10 different applications.
The presentation shows how we tackled the problem and what we learned about the efficient use of AI. To make better use of AI, we created a comprehensive Python program and used it to divide the migration project into several phases and these into many small steps. We have always given the LLM only relatively small tasks with sometimes extensive prompts, which has significantly improved the quality of its output. We then automatically generated the next prompt from these outputs and thus built up a process chain.
We will also show how much savings we have made compared to a purely manual migration and the amount of manual rework we could not avoid.
The insights and experiences gained serve as a template for further tool migrations in our company and therefore we want to share them here as well.
Key Takeaways:
Practical example for the benefits and limits of AI
Suggestions and ideas for AI-based tool migrations
Encouragement to start with AI-based projects
Talk Summary:
Supporting tool standardization, we migrate extensive commercial test automation frameworks to Java Playwright with the help of AI.
Our presentation shows our approach and learnings in using AI efficiently. We show how we do this with the help of a Python program and what efficiency gain we finally achieved.