Best Practices for Accelerating Delivery
summary
Claim that a $40,000 investment in visual modelling prevents $120,000 of defects and finds 6 of 30 requirements flawed (about 20%), demonstrated by converting wordy requirements into PlantUML/mermaid diagrams via ChatGPT and draw.io. An in‑house LLM, Planet Amplify self-healed test, patches a failing Codecept/Playwright script in CI and marks a warning so engineers are notified. Also builds a proposed Q model for lifecycle QE. Demos work end‑to‑end but still miss accessibility and messy DOM/ID issues that need human review.
AI-written summary · about
chapters (12)
- 0:00Accelerating delivery with AI
- 2:14Quality engineering challenges
- 3:39Shifting left quality
- 5:20AI requirements & testing
- 8:31Demo: AI state diagrams
- 12:46Testers validate AI models
- 16:19Testing as one lifecycle
- 17:45AI feedback loops in automation
- 20:59Demo: self-healing tests
- 27:21Feedback loops in CI/CD
- 30:57Modernizing QE with AI
- 33:39AI with human oversight
related talks
47mPanel Discussion: Agentic Engineering and Quality in Healthcare
06mLightning Talk: Automated Testing Done Right
45mPanel Discussion | How to deal with bad requirements in testing?
01h 21mPNSQC 2013 - Michael Mah
36mHow Testers Can Help Paving the Future of Modern Requirements Engineering
21mLowering End to End Testing Barriers
01h 04mBuilding on Success, Beyond the Obvious
44mExcellent Testing Starts With Excellent End-User Requirement Engineering and Specification