Welcome to the 322nd issue!

Something different this week.

Meet Playwright Quest — an online game that helps you learn Playwright, step-by-step.

I had fun playing it, and it works whether you're getting started or want to test your knowledge.

It's also been warmly welcomed by the testing community on Reddit.

So, happy learning! 🙂

Dawid Dylowicz

An interesting point of view from Kek Wee Tan who finds out that most bugs aren't in the model but in the instructions. For example, a guardrail that fixes one problem can potentially create another.

Moreover, Janna Loeffler points out why QA should build agents that make everyone else's work safer.

Are QA teams being quietly downsized because of AI? Gaurav Singh writes an honest, personal take on what's happening, why leaders fall for it and how testers should respond.

Also, Jaclyn Leigh has a good message to testers: You're Probably Doing More Than "Manual Testing".

While automation (and AI) can cover most regression paths, Jamin Chaput says they'll never replace the experience and empathy a skilled manual tester brings to the table.

In relation to that, you can also check out two insightful Reddit threads: Do you still run manual tests? To what extent? and Using AI for manual testing.

Dan Ashby looks at why AI testing tools get adopted but aren't always used, and what engineering leaders can do to help their teams build enough trust to change how they work.

At the same time, Janet Gregory outlines who's really in charge of AI Accountability.

Want to know why exploratory testing is important? Stuart Thomas gives a thorough, practical breakdown, including how to run it alongside AI systems.

Additionally, Callum Akehurst-Ryan advises to Start with the risks, not the method.

When AI can generate code and tests at speed, someone still has to check whether it makes sense. Julia Pottinger walks through the five questions to ask before approving any AI-assisted change.

Moreover, Anil Singha wrote a high-level overview of How I Built a Scalable Release Confidence Platform with Playwright, CI/CD, and AI.

Looking for a solution for flaky tests? Sławomir Radzymiński walks through a practical example where an AI agent investigates the failure first and only opens a pull request when the evidence points to a test defect.

Also, Gleb Bahmutov explains why End-to-end tests need to decide what can be retried to avoid flake with examples.

If you're not sure where to put assertions in your Playwright tests, Anton Gulin has a simple rule: page objects own the waiting, tests own the verdict, followed by advice on Reusing One Page Object Method for Success and Failure Cases.

This is a solid intro to visual regression testing, describing what it catches, how to fit it into your CI/CD pipeline and how to avoid flaky screenshot tests.

Similarly, Irfan Mujagić shows an example of why Visual Diffing is a Change Detector. You Need a Correctness Detector.

An interesting read from S Chathuranga Jayasinghe who argues that UI tests break not because selectors move but because tests go through a scripted path instead of verifying a goal.

Good practical guide from Abdur Raffae Masood on diagnosing flaky Playwright tests, from DOM race conditions and CSS animations to parallel run resource issues, with code examples for each fix.

Also, Katsuya Oura explains Why using getByRole is better for AI agents.

This is interesting. Stefan Judis ran an experiment comparing MCP and CLI token usage for Playwright agents and found no meaningful difference today — quite a change in just a few months!

Speaking of that, Julia Pottinger lists MCPs Every QA Should Switch On, while Andrii Cheparskyi shares What Google's AI Tools Are Actually Worth to QA Right Now.

If your Playwright test suite is growing and setup code keeps repeating everywhere, Mohammad Faisal Khatri has a good practical guide on building custom fixtures to fix that. You can also learn How to use Authentication setup in Playwright TypeScript.

Playwright 1.62 brings isolated retries that separate resource problems from bugs. Pratik Patel walks through every new feature with examples.

Moreover, I also enjoyed Svetlana Tretjakova's take on Playwright: Why TypeScript and Not JavaScript? and Combining UI and API in Playwright tests.

Is your QA team sitting on years of test cases that no one can find? Sergei Krapivin makes a case for a small MCP server that does hybrid search over your existing test base and drafts new checklists with project context.

Inspired by the "Mindfulness in Plain English" book, Lina Zubyte shares a refreshing QA perspective on AI overload and the importance of pausing and questioning what really matters.

In just 11 minutes, Alex Khvastovich showcases using AI to write Playwright tests, including resilient locators, page objects and official Playwright test agents.

Yet another interesting talk. This time, Joe Colantonio invited Keith Klain for a 48-minute discussion on the future of the software tester's role.

Born a Tester... 😂

Thanks for reading!

If you enjoy this newsletter and find it helpful in becoming a better tester, please consider sharing it with others.

Dawid Dylowicz