From Manual Tester to Agentic Wizard: The One-Hour Framework Transformation
The Journey from Manual Tester to Agentic Automation Engineer just got a whole lot easier. This session shows the many perks and pitfalls of Advanced Vibe Coding in a Playwright project. intelligent code completion can feel like having a mind reader as a coding partner. The AI can predict exactly what you're want to write – from adding fancy console.log icons (because who doesn’t want their test failures to look stylish?) to creating visual test boundaries with outline boxes. This is nice until you discover Claude Code’s agentic mode.
The Critical Difference: Context vs. Code Completion
While in-editor AI excels at predicting your next line of code, the agentic approach understands your entire project ecosystem. The real breakthrough comes when implementing big changes like test execution – configuring parallel workers and identifying tests that needed to run in series due to interdependencies. The agentic AI can see these relationships across the entire codebase, something impossible with line-by-line code assistance.
The Sweet Spot for QA Teams
For QA engineers transitioning from manual to automated testing, there’s a balance between moving fast and building maintainable test suites. The agentic approach proved invaluable for complex configurations like GitHub Actions workflows and understanding how test dependencies ripple through your entire automation framework – moving us beyond “vibe coding” to systematic, scalable test automation.
The Hidden Gem - Troubleshooting
In the final section we will review how Agentic AI was able to resolve an obscure Playwright issue. The most unexpected begavior was that it tried multiple approaches until it hit upon a solution. And even documented what it learned in the ReadMe file for future supporters of the framework.
Paul Grossman
Paul Grossman is known in the industry as the Dark Arts Wizard of AI Driven Test Automation. He has been a Test Automation Champion and Evangelist for 20+ years. He began as a Software Developer Engineer in Test with Mercury WinRunner before working with QTP/UFT for 16 years. He has experience in numerous test automation tool sets including Playwright in TypeScript, Selenium with Java and WebdriverIO in JavaScript.
In the past, he has automated the testing of medical equipment, insurance policies and entertainment kiosks. He is currently an independent consultant for a Trucking company automating the regression suite of tests with low code automation platforms. This also included providing mocks-ups of feature enhancements often requested by customers. Prior to that, he was a Project Manager for a freight tracking company based in Ann Arbor handling a WebdriverIO suite comprised of 1200+ automated test cases running every night.
In 2023, he co-authored "Enhanced Test Automation with WebdriverIO: Unlocking the Superpowers of Hybrid Testing Frameworks" with Larry Goddard. This was based on his years of experience to help new SDETs avoid real-world pitfalls and challenges. More books with Playwright and Cypress are in the planning stages.
Paul has been a conference speaker since 2004 including ZappleTech's monthly meetups and Joe Colantonio's Test Guild every Halloween. These always include a live demos, sharing tips from his experiences, with a little humor and the occasional magic trick. When Paul is not spending time automating test cases, he is producing demo videos on his a fledgling TikTok account as well as his YouTube and TwiX accounts with thousands of followers.
He just started a live multi-stream channel on LinkedIn, Facebook and Twitch.
Paul's current passion is learning more about Agentic AI and self-healing tests
