Ask Me Anything about AI, automation and Shift Left

AI is radically changing software development - but what does this mean for testers? The role is not dying out, but the way we work has to change: Those who only work through test cases will have a hard time. At the same time, many projects are experiencing a new quality crisis due to AI-generated code, which is bringing critical questioning and strategic thinking back into focus. From the question of automation levels to dealing with flaky tests and test strategy: the next few months will show who sees software testing as confidence-building - and who will lose out.

Podcast Episode: Ask Me Anything about AI, automation and Shift Left

In this year-end solo episode, I take stock of 176 episodes, 228 guests and 122,000 downloads - and answer your most burning questions from the community. Will AI make the testing job redundant? As a manual tester, do I have to learn programming now? And why is 100 percent test automation nonsense in most projects? I share my honest assessment of these questions, tell you about my most embarrassing podcast fuck-up in 2024 and reveal which episodes made me think the most. There’s also an outlook on what’s waiting for you in the Software Testing podcast in 2025 - and why test strategies need to be back on the table right now.

“You don’t have to follow every trend.” - Richard Seidl

Highlights der Episode

• AI does not make testers superfluous, but the job is changing massively due to new development speeds.
• Isolate flaky tests immediately: They destroy trust in pipelines and need to be analyzed or deleted.
• Shift left starts in refinement, not in code review - this is where a common understanding of quality is created.
• 100% automation only works if the software architecture is designed for it from the outset.
• Do not follow every tool trend - Selenium remains valid if it is stable and fulfills its purpose.

Year in review 2025: Insights, mistakes and looking ahead

Review of one year of podcast software testing

2025 was an eventful year for the Software Testing podcast. Richie’s review is impressive: 176 German episodes, 228 guests and 300,000 downloads. This is a remarkable achievement for a niche podcast, especially in German-speaking countries. A sign that the topic of software quality moves many people. The community is growing and the podcast now has around 3,000 subscribers.

A quick look at the figures reveals some exciting facts. A quarter of listeners use Apple Podcasts or Spotify. The others are spread across various players. Even the videos on YouTube are gaining popularity.

But for Richie, one thing is clear: the podcast thrives on exchange. Feedback, questions and criticism are important in order to improve the format. This is also reflected in the course of this year.

Listener questions: What’s on the testing community’s mind at the moment?

At the end of the year, Richie collected and answered some questions from the community. Many of them revolve around uncertainties and trends.

Will AI make the testing job redundant?

Two years ago, Richie might have said yes. He now has a more nuanced view. Testing is changing, yes. AI is creating new dynamics, but testers are needed. Machines cannot completely replace critical thinking, strategy development, communication and building trust. Rather, the tasks are growing: Precisely because AI generates a lot of code, it needs a scrutinizing eye. Quality is getting more attention again.

Do manual testers now have to learn to program?

Not necessarily. Technical understanding helps, but hardcore programming is not a must. What is more important is an understanding of processes, quality improvement and the clever use of methods. Dealing with AI in particular is becoming a relevant skill. Scripting can help - but it doesn’t have to be the way for everyone.

Tool changes and automation:

Should you switch from Selenium to Playwright because everyone is doing it right now? Richie says: Don’t chase after every hype. If the current tool is stable, there is hardly any reason to switch unless you see clear advantages such as a better connection to development or availability of expertise in the team. Switching is only worthwhile with a clear objective.

100% automation sounds fancy, but it often doesn’t make sense. Many projects come up against technical and organizational limits. The focus should be on meaningful automation, not on quantity. And yes: good test pyramids are based on different test types, not just end-to-end.

Flaky tests - delete or investigate?

Flaky tests destroy trust. Get rid of them from the pipelines, then analyze why they are unstable. Perhaps they can be stabilized or need to be removed completely. Goal: Stable and meaningful test systems.

Errors are part of this: Dealing openly with breakdowns

An honest look behind the scenes is also part of this year’s conclusion. Richie freely reports how he forgot to test the guest microphone during conference recordings. The result: unusable recordings, a lot of trouble and a need to catch up. But mistakes happen. It is important to learn from them and improve processes. Openness helps, even in day-to-day testing: it’s better to be honest about problems than to cover them up.

Favorite episodes and special moments

A podcast thrives on diversity and exchange. Many conversations were special for Richie this year. For example, episodes about enterprise testing, microservices, the new professional fields of autonomics and the use of AI for legacy modernization. The mixture of technical, methodological and human topics is what makes it so appealing to him.

The community is invited to share their own favorite episodes and requests - feedback helps with the planning of future episodes.

Outlook for 2026: Growth and new topics

Richie is planning content growth for the coming year. The international podcast “Software Testing Unleashed” is to be further established. New topics such as soft skills, conflict management and transfer services in the testing world are on the agenda. The classic test strategy is also becoming more important again in times of AI: prioritization and the right focus on relevant quality characteristics are becoming more important.