github
BIO Piotr Szczepaniuk
“Most AI QA demos look impressive…
but they don’t work in production.”

CURRENT STATE
What we see today
Reality

REAL PROBLEM
Why this fails

Cannot go to CI/CD
Real issue


ROOT CAUSE

“We don’t control what AI produces.”

SOLUTION

HUMAN + AI MODEL
ARCHITECTURE



AI QA Agent Workflow

MCP vs PLAYWRIGHT-CLI vs SKILLS

Demo
Real Use Cases (From Demo to Production)
  • Smoke testing applications
    → verify critical paths (login, checkout, API)
  • Post-deploy validation
    → run tests after release, detect regressions
  • Bug reproduction
    → replay flows with traces, video, screenshots
  • AI QA agents in production
    → generate once → execute many times → validate results


FUTURE
“QA Engineer becomes
AI Orchestrator & Validator”
What changes
What stays
KEY TAKEAWAYS
  • AI alone is not enough
  • architecture matters more than tools
  • MCP ≠ execution
  • Playwright CLI = stability
  • skills = scalability
  • validation is critical
TO BE CONTINUED…

“How AI can help find hidden bugs — not just generate regression tests”
Q&A: Comments & Discussion
Now it's time for discussion.
If you have any questions or comments about Agentic AI Workflows for QA, feel free to ask them now — or reach out to me afterward.

loveautomate.com

LoveAutomate – AI-Driven QA Workflows

AI-powered agentic workflows for QA and test automation.

Made with