Welcome to the 317th issue!

We all use AI now, but are we doing it right?

I liked the recently updated version of How I Use AI to Code by Chris Parsons.

There's some great practical advice for engineers, including strong references to other sources.

On top of that, I also recommend a discussion — How do you use Claude Code for QA?

Happy testing! 🙂

Dawid Dylowicz

Now that most people switch to building with agents, you may start hearing more about SDD. Lisa Crispin and Janet Gregory explain what it is and how it fits the Holistic Testing Model.

Similarly, Dmytro Stasyuk shares When Specifications Need More Than Prose.

DORA metrics defined the last decade as the "go-to" indicators of how systems perform. But what changes in the age of AI? Callum Akehurst-Ryan shares some good thoughts.

Shipping faster doesn't necessarily mean shipping better. Jaclyn Leigh shares why senior quality roles are growing in an environment where AI-generated tests still miss edge cases.

Similarly, Jitesh Gosai asks — If AI helps teams build faster, who helps them build better?

Want to trust your AI testing agent? Yiğit Taş explains why passing tests is just half of the story and shows three traps that lead to fake coverage and made-up expectations.

Similarly, Mehmet Serhat Özdursun describes The True Cost of "Autonomous" QA Agents (And What the Viral Diagrams Ignore).

Good read from Dennis Byrd on how AI changed the quality for them. Apparently, even non-engineers now build tools while over 40% of changes are auto-approved.

At the same time, Keith Klain concisely outlines why Testing is Beautiful (well, at least until it's done properly).

Stuart Thomas explains why it's important to start with AI-assisted code review and test scaffolding to build trust for AI to write production code.

Similarly, Sourojit Das says that AI Can Write Tests. It Still Can't Generate Trust, while Christopher Meiklejohn shares an interesting story where The Test Suite Was the Incident.

Chathuranga Siriwardhana describes how they grew code coverage from 22% to 75% while making end-to-end tests a strict release gate.

Does testing an API endpoint tell you the whole feature works? Gil Zilberfeld argues it doesn't and explains why you need multi-step, workflow-level tests instead.

A good reminder from Kevin Roe that the test pyramid is more about a relationship than just a diagram, encouraging to read unit tests, pair on integration work and which lower layer could have caught a given bug.

Moreover, Daniel Brown explains in detail Why you should be testing your architecture using ArchUnit.

If your test framework locator strategy falls back through id, aria-label, name, CSS, then XPath, you might be testing the wrong button. Ricardo Franco shows how static analysis can help you spot these before any test runs.

Wondering what load testing tool to choose? NaveenKumar Namachivayam compares JMeter, k6 and Locust, including scripting, concurrency, CI fit and the shortcomings.

Earlier on, NaveenKumar also described Why You Should Use Gaussian Timer in JMeter, k6, and Locust.

Want to use Playwright with AI without burning your budget? David Mello shares lessons from StarEast 2026 on why the Playwright CLI uses far fewer tokens than the MCP server.

Furthermore, Varsha Rajput shares How To Configure Multiple Environments in Playwright.

Sure, AI can quickly write tens of Playwright tests but then some of them might be flaky. Anton Gulin shows how to review them against 10 solid rules to keep your suite stable.

At the same time, Arpita Biswas wonders: If Playwright Has Auto-Waiting, Why Do We Still Need Explicit Waits?

Filip Koss put together a long, friendly guide to testing in Python with pytest. It walks through fixtures, parametrisation, mocking, and the unit-integration-E2E split with small, easy examples.

Also, Irfan Mujagić wrote a good overview of Quarantining Tests: A pytest Pattern for Failing Tests Without Breaking CI.

Automating pages with reCAPTCHA protection can be tricky. In this helpful, 12-minute video, Karthik KK shows a workaround using open-source Playwright and Browserless tools.

If only... 😅

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