ABOUT THIS EPISODE
With Klaudia Dussa Zieger I talk about where language models actually deliver in testing and where they fall flat. We get into the concrete details: why a small function like generating keyword documentation works so well with a standard chat model, why reviews only started working once they switched to a reasoning model, and why one engineer working on embedded legacy code said no to AI help without hesitation. Klaudia also shares the translation use case I found most vivid, turning a cryptic technical defect message into something a domain tester can actually read.
"The more autonomy you give to an AI agent, the better it has to be." - Klaudia Dussa Zieger
Klaudia Dussa Zieger has been working in the field of software testing and quality assurance for more than 25 years. She is the team leader responsible for consulting at imbus AG. She is particularly interested in test management, the continuous improvement of the test process and the professional training and further education of testers and has been a trainer for the ISTQB Certified Tester Foundation and Advanced Level as well as a lecturer for software testing at the University of Erlangen-Nuremberg for more than 20 years. Since March 2009, Klaudia has been chairwoman of the DIN working committee on systems and software engineering and is actively involved in the development of standards at international level.
Highlights:
- Reasoning models outperform standard chat models for code review tasks: switching from GPT-4o to O1 was what made automated keyword review produce usable results.
- Small, tightly scoped AI functions deliver faster adoption than large end-to-end automation, because testers can verify the output and stay in control of the process.
- Self-healing test automation that uses embeddings to repair broken locators risks masking real defects, because the system cannot distinguish an accidental change from an intentional one.
- As AI agents gain autonomy and operate without a human between each step, the quality requirements for their output rise sharply, which closes the gap between using AI and testing AI.
📘 Tool migration ahead? Teamscale playbook: migrate only 20% of tests, keep nearly all bug detection. Download for free
