Welcome to the 307th issue!

AI is the hottest topic right now. And for a reason!

So today, I'm bringing to you two big discussions among the testing community:

• Do you use any tools or AI for testing?
• AI use cases for QA

As a bonus, I also recommend Chris Kenst's take on My AI Tooling Stack: The March 2026 Edition.

Happy testing!

Dawid Dylowicz

Good question. Axel Kirchner makes the case that AI-assisted development doesn't reduce the need for quality assurance but changes where that work needs to happen.

Similarly, Irfan Mujagić also asks: Who Tests the Tests? AI, QA, and the Verification Paradox.

Karthik Sivakumar argues that the biggest drag on engineering velocity isn't bad code — it's poorly designed test infrastructure.

What's more, Betty Lin shares about What AI Still Cannot See in Software Quality (Part 1): Requirement Conflicts, followed by parts two, three and four.

Not all parts of test automation benefit from AI in the same way. Viktar Karanevich points to a few areas and shares an SDET transformation roadmap.

Wondering how to keep up with the changing world of software? Ben Fellows sees the tester role changing too and suggests what skills will be the most important.

Additionally, Neil Duggan points out why Your QA Team in 2028 Doesn't Look Like Your QA Team Today.

This is good advice from Callum Akehurst-Ryan on how QA teams should adapt their approach when developers are leaning heavily on AI assistance.

And there's a reason why Dan Belcher announces we live in The Most Exciting Moment in Software History due to the recent advancement of LLM models.

Preparing for your next QA role? Bartosz Kustra breaks down what actually matters to interviewers in the current market.

Similarly, Joshua Bihun also advises on Hiring QA Engineers: Start With What You Actually Need.

This is a good walkthrough from Martín Marchetto on how to structure test code so that both humans and AI agents can work with it.

Accessibility test reports can be noisy, but Benjamin Bischoff and the Trivago team found a better, more actionable way of dealing with it.

Pramod Dutta ran an interesting experiment, observing manual testers implement test suites much faster without a traditional learning path, thanks to AI.

Also, Megan Ozanne wrote a good article about Coaching with code: using AI to learn how to automate tests.

David Mello presents a tool that generates API test cases directly from an existing contract and proves to be surprisingly good at covering edge cases.

An interesting idea from Mehmet Serhat Özdursun, suggesting using AI to analyse git diffs and give testers a structured view of where a change is most likely to cause problems.

Skills have become the default way to provide context and instructions to AI agents. But how to do it right? Angie Jones shares some tips.

Similarly, Gurudatt S A advises to Stop Re-Explaining Your Test Conventions to Claude — Use SKILL.md and Deep Shah explains why I Stopped Writing Prompts Every Day.

Swati Sabharwal spent six months figuring out how to make AI coding assistance work in a real team environment and shares her insights.

Speaking of that, James Kip explains why Your Linter Can't Save You: Using .github/copilot-instructions.md to Catch What Static Analysis Misses.

Emre Erkek describes in detail how they implemented an end-to-end mobile automation with Maestro across multiple business domains.

Thinking about switching to Playwright? Asjad Khan wrote an extensive overview on how to do it right, especially about the challenging parts.

In this 10-minute overview, Karthik KK demonstrates a custom-built autonomous UI testing agent with Playwright, inspired by autoresearch's approach.

Playwright CLI was released just a few weeks ago and it's now the default interface for AI agents. If you want to see it in action, Lucas Smit recorded a helpful 18-minute demo.

Bugs in 2026... 😅

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