Library › Book 1 › Future Directions
Future Directions
25.1🔒Keeping Score: How the Last Predictions FaredA futures chapter you cannot grade is astrology, so grade the old predictions first. The January 2026 edition made four "near-term" calls:
25.2🔒Already Arrived (2026 -- No Longer "Future")First-party test agents. Playwright's planner / generator / healer shipped in v1.56 and matured through 1.61; the plan -> generate -> heal…
25.3🔒Near-Term (2026-2027)What: A persistent model of the application -- pages, flows, elements -- that survives across agent sessions.
25.4🔒Medium-Term (2027-2028)Autonomous test generation at suite scale. The agent explores the application and grows the suite with minimal human scenario-writing…
25.5🔒Long-Term (2028+)Self-writing test suites. The AI continuously monitors the application, detects changes, generates and runs tests, and reports issues…
25.6🔒Industry Convergence PointsProtocol: BiDi won the argument.
25.7🔒What This Means for Your Career1. AI agent architectures -- ReAct patterns, multi-agent systems, tool use. 2. Testing AI systems themselves -- now explicit in mid-2026 QA…
25.8🔒Key Takeaways- Grade your predictions: test agents, AI locators, and BiDi networking shipped within six months; persistent app memory did not. - The…
25.9🔒Q&ASelf-Assessment Quiz1. Which January 2026 "near-term" predictions had shipped by July 2026, and which had not? 2. Why is persistent application memory the…
25.10🔒Exercises[Beginner] Exercise 25.1: List three ways your testing workflow would change with persistent application memory. Which saves the most…
25.11🔒Career Translation- Maintained a graded prediction scorecard for AI-testing trends, correctly identifying first-party test agents and AI locators as…
25.12🔒Q&AInterview Depth CheckPrompt: Your VP asks for a 12-month AI-test-automation roadmap. The team runs traditional Playwright with no agent involvement.
25.13🔒Project OverviewBuild a production-ready framework demonstrating the book's major concepts on the July 2026 reference stack: the Playwright CLI skill for…
25.14🔒Phase 1: Project Setup
25.15🔒Phase 2: Scaffold the StackYour coding agent now knows the browser vocabulary (SKILL.md) and has planner/generator/healer definitions available.
25.16🔒Phase 3: Write the PlansWrite specs/auth.md in natural language -- the plan is the artifact humans review:
25.17🔒Phase 4: Seed Test and GenerationWrite tests/seed.spec.ts so agents inherit your setup:
25.18🔒Phase 5: The CLI Smoke LayerWrite scripts/run-smoke.sh -- a fast, cheap layer exercising the CLI skill directly, one named session per test, authenticated state saved…
25.19🔒Phase 6: CI/CD PipelineWrite .github/workflows/test.yml:
25.20🔒Phase 7: Reporting- The generated suite emits JUnit XML via Playwright's built-in reporter (reporter: [['junit', { outputFile: 'results.xml' }]]). - Extend…
25.21🔒Evaluation CriteriaYour capstone project should demonstrate:
25.22🔒Stretch Goals- Add visual regression: screenshot to disk each run, compare against a committed baseline - Measure pass^k: run the generated suite five…