Elevating Quality: How AI Enhances Testing in Continuous Delivery Pipelines
summary
Tilaa describes building an AI pipeline that finds end-to-end coverage gaps by parsing Vue code back from Apollo calls, emitting YAML coverage format, and caching with checksums to avoid repeated work. The team uses a local model on GPUs plus Bitlane AST tracing to reduce context window issues, wraps the logic in the Quinton AI agent, and reports a jump from 75% to 90+% confidence in test coverage; Quinton runs manually today and is not yet in CI.
AI-written summary · about
chapters (19)
- 0:00AI-enhanced testing intro
- 1:27Why testing is hard
- 2:51Testing pyramid limits
- 5:42Modern development principles at Tila
- 6:46Deployment challenges
- 8:34Bug reveals testing gaps
- 10:22Mapping features manually
- 12:08Quinton AI agent intro
- 13:54AI tool challenges
- 14:36Defining a feature
- 16:00Context window problem
- 17:26Bitlane for tokens
- 19:11Making AI outputs deterministic
- 20:36YAML for features
- 21:43Risk-based test filtering
- 23:05Caching for performance
- 24:51Quinton results
- 25:32Lessons learned
- 26:56Final thoughts
related talks
46mQA to QE: Scaling Quality with Ownership, Code, and Culture
26mAccelerating DevOps with AI
26mAI and the Future of Testing
47mPanel Discussion: Who Tests the Machines? QE Leaders on Quality in the Age of AI-Written Code
25mRethinking Software Quality: Why API Testing Needs to Change
29mQA vs AI: Friends or Foes?
16mTransforming Test Automation with AI and Machine Learning
26mMaximizing Test Coverage with AI-Powered Test Automation