Workshop: The Full Agentic QA Loop for Web Applications

on YouTube ↗53m

" summary

Workshop shows an end-to-end QA pipeline that turns a PRD into plain-English tests, runs them in a real browser, and emits portable proof. The demo uses kane-cli to extract four use cases, generate 13 acceptance criteria, author four runnable tests, capture DOM/actions/network/console into an evidence pack, and report coverage (one run shows 12 of 15 coverage). The CLI replays runs deterministically, supports evolve/maintain commands, and offers 10,000 free credits for hands-on setup.

AI-written summary · about

" chapters (42)

1. 0:00Start
2. 0:37Speaker Intro - Siddhant Sinha, Lead Member of Technical Staff at TestMu AI
3. 2:10AI Writes Code in Seconds - So Who Tests the Whole Web Flow?
4. 2:56The Same AI That Wrote the Code Also Writes Its Test
5. 3:11Brittle, Slow Tests Never Proven in a Real Browser
6. 4:12The Second Problem: PMs, Devs and QAs Working From One PRD
7. 5:29Drift, Stale Tests and Coverage Numbers That Are Really Guesses
8. 6:01One Loop, One CLI: PRD In; Criteria, Scenarios and Tests Out
9. 6:32Tests Authored in Plain English, Replayed in CI With Evidence as Proof
10. 7:33Claiming 10,000 Free Credits to Follow Along
11. 8:18Installing the CLI and Logging In to TestMu AI
12. 8:34Test Manager: Where Authored Tests and Runs Are Audited
13. 11:08The Sample GitHub Repo and the Five Things the Workshop Covers
14. 12:24What Context Means Here, and Maintaining It as the PRD Changes
15. 13:26Running context ingest on the PRD
16. 13:58The Demo App: A Local V16 Food Ordering Website
17. 14:59Four Use Cases Extracted From the PRD
18. 16:30Changing a Use Case's Risk Level in Plain English
19. 17:47context list and context view: The Graph of Linked Use Cases
20. 19:34Designing Criteria and Scenarios Before Tests
21. 20:20Running design tests on Use Case One
22. 20:52The Review Gate: review and approve Commands
23. 21:53Sessions Let You Resume Instead of Designing Everything at Once
24. 22:09Thirteen Acceptance Criteria From a Single Use Case
25. 23:25Not Just UI - Asserting on Network and Console Too
26. 24:57The Agent Asks Two Questions Before Building Scenarios
27. 25:43Choosing a Scenario Budget - and Why No Limit Avoids Gaps
28. 26:46Four Scenarios: Two Happy, One Negative, One Boundary
29. 28:17Wiring Scenarios to Acceptance Criteria
30. 31:04Four Tests Created, Each Verifying a Set of Criteria
31. 32:52The Graph View, Lineage and the maintain evolve Command
32. 34:55Opening the Plain-English test.md in VS Code
33. 37:12Listing Tests and Running One in the Browser
34. 38:29Behind the Scenes: The Agent Reads Screen, Tree, Network and Console
35. 39:30After the First Pass, Execution Is Just Replay
36. 41:20The Run Completes: Artifacts, Version History and Test Manager Link
37. 43:25Re-Running the Authored Test as a Fast Replay
38. 44:27The Evidence Pack - Served Locally, Shared via evidence.lambdatest.com
39. 45:13Coverage: 12 of 15 Acceptance Criteria From One Test
40. 48:18Q&A: How Agentic Review Detects That a Test Has Drifted
41. 49:36Q&A: Preventing an Updated Test From Masking a Real Bug
42. 50:54Q&A: Who Is Responsible When an AI Agent Causes a Security Incident

" related talks

• 28mTesting at the Speed of Thought: QA in the Era of GenAI appsRobin, Shriyansh · Agile India 2025
• 35mState of end-to-end testing for modern web applicationsVlad Filippov · REFACTOR DX 2023
• 43mLean QA: Automation, Monitoring, ExperimentsKateryna Chernikova · Selenium Camp 2019
• 46mThe AI Revolution In TestingVimukthi Saranga · Spartans Summit 2025
• 56mNo QA resources? You can still deliver a quality websiteSadia Rodriguez · MidCamp 2020
• 46mQA to QE: Scaling Quality with Ownership, Code, and CultureNatalie Cole · TestMu Conference 2025
• 50mAutomate Finding & Fixing Bugs with Replay QABrian Hackett · Vibe Coder Conference 2026
• 10mFrom PRDs to test cases: building AI-powered QAManas Chaturvedi · The Fifth Elephant Pune edition 2026