Meaghan Osagie (Lewis)

One reaction I keep seeing in QA (and engineering more broadly) to “the AI thing” is: “I’ve been around for all these changes. People have been saying QA is dead forever. We’re still here.” Just to be clear: I’m not saying QA is dead. I don’t think many serious people are actually claiming that. This reaction feels more like a knee-jerk defense than a real analysis. What’s challenging for me is that when you step back and actually look at the chain of change, this isn’t “just another iteration of the same tools.” For example: We can now write automated tests that don’t rely on the DOM, aren’t flaky, are deterministic, and get cheaper/faster over time. That alone breaks several of the historical constraints that made test automation hard. Within a few weeks, we transitioned a team that had been doing manual UAT into one where those same engineers were scripting automated tests before tickets were even ready, then essentially just hitting “play” when the feature landed. This was a team that had tried (and failed) multiple automation initiatives in the past because delivery pace always outstripped their ability to automate. Suddenly they had 200+ smoke and sanity tests built in a matter of days. I don’t know how you look at that and say, “Yeah, this is the same as before.” Part of this feels like how magazine layout professionals reacted for decades. Printing technology kept evolving, but the core process stayed the same: humans laying out pages by hand. Then computers showed up. Sure, you were still “laying out a magazine,” but it became a fundamentally different process with a different set of skills. Many people made the transition. The ones who said “this kills the art” or “this is a fad” mostly didn’t. This isn’t about QA disappearing. It’s about the work changing shape. And history suggests that when multiple constraints collapse at once, pretending it’s business as usual is the riskiest reaction of all.

17

7 Comments

We built the BrowserStack Website Scanner AI Agents to act as an extension of your QE team. Think of them as pair testers that never get tired and have memorized every WCAG guideline. We are launching with three core capabilities: Visual Review Agent: Drastically reduces review time by highlighting meaningful UI changes. Accessibility Issue Detection Agent: Identifies complex context that automated tools usually miss. Accessibility Remediation Agent: Provides expert-level guidance and fixes instantly. This is how we scale quality. Check it out: https://lnkd.in/gTYNMwBj

52

Here is Part 2 of my series on #vibecoding a pipeline. This time it is building up synthetic tests for production monitoring. The natural candidate for this was to use Checkly cos it is #Playwright native so transitioning existing tests to become monitors should be straightforward (note: should be). As with anything in software, it's never straightforward (details in the article). Many thanks for Hannes Lenke and team for reviewing it - I know they are working to make creating and maintaining synthetic tests as seamless as possible in this new vibe coding world. #AI #softwareengineering #softwarequality #devops #syntheticmonitoring https://lnkd.in/gdChwY8G

19

How do you break into QA without a traditional tech background? 🤔 ☕ Judy Mosley Mosley did it—and she’s sharing exactly how. In this episode of the TestGuild Automation podcast, I chat with Judy about her pivot from full-time mom to respected QA engineer at TextUs. Spoiler: it didn’t involve a CS degree or a perfect résumé. Instead, Judy leaned into: ✅ A development bootcamp that sparked her passion ✅ The value of “breaking things” in QA ✅ Blogging to build credibility and community ✅ Tackling imposter syndrome head-on ✅ Growing as an introvert in a noisy industry ✅ Using both manual and automation testing to stay agile ✅ And navigating how AI is reshaping QA careers Whether you’re switching fields, starting from scratch, or just need fresh motivation—this one’s worth a listen. 🎧 Tune in: https://lnkd.in/dqeiaXqw #SoftwareTesting #QACareers #TestAutomation #WomenInTech #CareerChange #IntrovertsInTech #TechCareers

29

6 Comments

Did we make a mistake pivoting away from dedicated QA folks in software? Admittedly, my teams have done OK without them, but now I'm starting to see value that I didn't before. There are a lot of obvious wins for QA people, not the least of which is that they almost universally have an innate sense of curiosity that makes them good at breaking things. It's not even just exploratory UI where this applies, because the good ones can do the same with API's. What I didn't see before is that they also keep developers and product managers honest by making sure that they develop stories that are actually testable. This means writing down clear outcomes and acceptance criteria. This is so important, because engineers are prone to interpreting less defined things on their own, or letting scope grow unchecked. Good AC means everyone agrees about where to go, making coding and testing a clear endeavor. Without QA, I think we slip into a mode of what we think is good enough. When there's no separate person to only test, no one is there to say, "That story is not fully baked."

635

204 Comments

Visual bugs are silently killing your user experience. Your functional tests pass, but your users see misaligned buttons, broken layouts, and styling inconsistencies that undermine brand trust. The reality? Most QA teams catch visual regressions only after they hit production. Ready to move beyond "it works" to "it looks perfect"? 👇 This comprehensive guide breaks down exactly how enterprise teams can adopt visual testing without overwhelming their existing workflows. https://lnkd.in/gixkNxSm #QualityAssurance #VisualTesting #SoftwareTesting #DevOps #UserExperience

25

Stop trying to write better prompts. Prompt engineering is a transition phase. Loop Engineering is the endgame. If you are just vibing with a chat interface, you aren't an engineer—you're a glorified QA tester for an LLM. The engineers of this decade won't be coders; they will be System Designers who build the constraints that keep agents on the rails. Are you building loops, or are you just vibing? #SoftwareEngineering #AI #AgenticAI #GoogleTech #FutureOfWork

57

9 Comments

“I need to see what the AI is doing.” That’s the most common refrain we hear from QA and engineering leaders exploring AI-driven testing. It’s not resistance to automation—it’s a demand for visibility. It’s not helpful to have a black box deciding what to test, how to test it, or what counts as a pass or fail. As one engineer put it, “If I can’t read the code itself, I don’t really know what’s happening under the hood.” Another said bluntly, “When the tool hides feedback, it just creates more pain.” There’s also a deeper concern: “If AI goes down the wrong path and people have become over-reliant, they forget how to fix it.” When systems become opaque, reliability erodes, and so do human skills. The future of QA calls for AI tools that deliver deterministic, interpretable code you can read, logic you can trace, and results you can defend. AI should accelerate testing, not obscure it. The moment visibility disappears, so does trust. To read more on codeless AI tools, click the link in the comment below #ai #aitesting #softwaretesting #softwaredevelopment #cto

31

2 Comments

🚨 QA Isn’t Dead - It Just Got Prompted. Everyone's rushing to build with LLMs. Few are testing them properly. Fewer realize how badly they’re breaking silently. Here’s the truth 👇 🔍 Traditional QA doesn’t cut it anymore. When you're dealing with prompts instead of functions, and stochastic models instead of deterministic code… You’re not testing code paths. You’re testing language behavior. Welcome to the world of Prompt Testing and LLM Validation. ❓Why should QA care? Because that fancy “AI Assistant” you shipped last week: Returns wrong info 17% of the time 🧨 Refuses basic instructions if phrased slightly differently 🤷‍♂️ Leaks PII when asked nicely 😬 And no one’s writing Selenium tests for that. ⚙️ What we do now: ✅ Regression prompt suites - same intent, 20+ phrasings ✅ Output validation - not just does it respond, but does it comply? ✅ LLM hallucination checkers -automated fact extraction + source diff ✅ Red teaming via prompt stress tests -think pen-testing, but with words ✅ Evaluation metrics beyond BLEU - coherence, helpfulness, faithfulness (and human evals if you're brave) 🛡️ QA in the LLM Era is Language-First. This isn’t about chasing bugs. It’s about guarding trust. Because when your AI makes a mistake, it’s not a 500 error - it’s a confident lie. And your users won’t file bug reports. They’ll just leave. QA teams: it’s time to evolve. Prompt testing isn’t a “nice to have.” It’s the new smoke test. #AIQA #PromptEngineering #LLMTesting hashtag #QualityAssurance #ChatGPT #OpenAI #Automation #ProductQuality #QA

26

1 Comment

Senior engineers don’t panic when requirements are fuzzy. Mid-level: “This story isn’t clear. I can’t start.” Senior: “This isn’t clear. Let’s clarify together.” Seniors: 🔹 Call out ambiguity early 🔹 Break problems into smaller, testable steps 🔹 Collaborate with product/design to shape the work Ambiguity isn’t a blocker - it’s an invitation to lead.

10

1 Comment

Behavioral interviews: A QA engineer’s guide As the founder of Codemify, I’ve seen countless talented engineers struggle with one critical hurdle: behavioral interviews. Technical skills get you in the door, but soft skills seal the deal. We held a mock interview session that revealed eye-opening gaps — and proven solutions. Here’s what every QA professional should know: The problem: Many candidates focus solely on technical prowess, forgetting that companies hire people, not just skill sets. Behavioral questions assess: How you collaborate (e.g., resolving team conflicts) How you grow (e.g., handling failure or knowledge gaps) How you drive impact (e.g., metrics-driven results) The fix: 1️⃣ “Tell me about yourself” → This isn’t a resume recap. Share why your work mattered. Example: “Led test automation for a healthcare app, reducing regression time by 40%—which directly accelerated patient portal updates.” 2️⃣ Strengths → Tie them to team success. Instead of “I’m detail-oriented,” say: “My focus on edge-case testing caught 15+ critical bugs pre-launch, saving 50+ dev hours.” 3️⃣ Weaknesses → Be strategic. Discuss skills you’re actively improving, like: “I’m deepening my mobile automation expertise via Sauce Labs certifications.” 4️⃣ STAR Method → Structure stories with Situation, Task, Action, Result. Numbers matter! ____ At Codemify, we don’t just teach tools — we prepare you for the real hiring process. Our goal is to bridge the gap between technical ability and career success. That’s why our programs include: HR-mocked interviews Engineering leadership coaching (to articulate your impact) Project-based learning (to build measurable achievements) For those ready to level up: Our Automation QA IntroWeek starts May 3. Visit https://lnkd.in/gQCesFcN to reserve your spot. Full session is available here: https://lnkd.in/gVbYveCe #QualityEngineering #CareerGrowth #TechEducation #Codemify

17

2 Comments

We've been in the testing space for 14 years. And there's one problem that's never gone away. QA teams tell us they spend up to 40% of their time just writing test cases. Not testing. Not finding bugs. Just documenting scenarios. It's the least interesting part of their job, but it takes the most time. I kept thinking, why hasn't this been solved? Generic AI tools generate test cases, sure. But they miss edge cases. They don't understand context. Teams end up rewriting everything anyway. So we asked: what if an AI actually understood testing? We built the BrowserStack Test Case Generator Agent. It's built on years of testing expertise, understanding what good test cases look like and how real teams work. Early users are creating test cases 90% faster. But here's what matters more, they're finally doing the work they signed up for. Finding real issues. Improving products. That's the shift I care about. Not just speed. But giving people their time back to do work that actually matters. Would love to hear what you think: https://lnkd.in/dfD8_a7F

353

10 Comments

I am glad folks at Cypress.io read my blog post "Pick E2E Tests To Run Using AI Summaries" and now are showing test summaries on dashboard. Next thing they will probably suggest tests to run. You can easily derive these text descriptions. https://lnkd.in/gUh2Srtm 

44

4 Comments

I usually post about how important QA is after a major outage like there was this week. But that’s becoming so frequent I don’t want to bother constructing an entire post. So, here’s the main points: Stop laying people off, especially when profits are healthy. The workload is too much for many teams, morale plummets and things slip through. Stop looking at QA as an expense. It is a necessary part of development. One builds, one tests, result is a more reliable product. You can be a scion in tech just to become a meme laughingstock overnight. Also, entrepreneurs: it’s a great time to make better choices than those that are established and not protecting their teams and products. Reliabity matters. Guess I made a post after all. Sure it won’t be the last. #softwareengineering #qualitymatters #outages #layoffs #techleadership #entrepeneurs #startups #qa #engineeringleadership #productreliability

30

6 Comments

🚨 Why You Should Run Regression Tests—Even After Changing Just a Comma In agile development, speed matters—but stability matters more. This blog breaks down regression testing: what it really means, why it’s unrelated to priority/severity, and how it acts as your final safety net before release. Changed code? Changed build. Whether you’re fixing a critical bug or adjusting a comma, regression testing is non-negotiable. Here’s what I’ll cover: ✅ The true purpose of regression testing 🔄 Its role across rapid releases 🚦 Why it’s essential for every change (P0 to P3) 🔧 How build resets demand regression resets 📖 Read the blog to understand why regression testing is your QA safety net

35

2 Comments