Case study
Nzila
Feature Intelligence & Execution Replay
Unifying automated tests, frontend runtime behavior, API traces, and CI context into feature-level execution intelligence—with visual replay and human-readable outcomes.
Problem
Teams see whether tests pass or fail, but not whether a feature actually worked as a real user experience—or where and why execution broke across UI, API, and runtime.
Context
CI reports outcomes; observability tools surface logs and metrics; session replay targets production users. None of them connect test results, React execution, backend traces, and user impact at the feature level.
Solution
Building a platform that reconstructs complete feature execution flows: Vitest/Jest/Playwright results, component and interaction traces, API lifecycle, runtime signals, and pipeline context in one timeline. Execution Replay visualizes rendering, inputs, requests, state transitions, and failure points—plus a Business Mode that translates technical runs into plain-language explanations for non-engineers.
Architecture
- Feature-level execution graph linking tests → components → API calls → outcomes
- Replay UI: timeline, interaction track, and failure anchoring for debugging
- Dual presentation layer: Developer Mode (detailed sequencing) vs. Business Mode (narrative summary)
- CLI-first onboarding (`npx`) with auto-detect setup, SDK install, and runner configuration
- Ingestion pipeline for CI/CD context and runtime events alongside test artifacts
Technical challenges
- Correlating heterogeneous signals (test runners, React runtime, APIs) without overwhelming the UI
- Keeping replay performant and readable for long execution paths
- Generating trustworthy human summaries from technical execution data
- Zero-config DX while supporting varied project layouts and auth flows
Results
- — Shift from pass/fail reporting to “where and why” the feature broke in user terms
- — Faster regression triage via visual execution instead of log archaeology alone
- — Bridge between engineering detail and stakeholder communication (Business Mode)
- — Foundation for correlating tests with real runtime behavior in one product surface
Lessons learned
- — Feature intelligence is a UX problem as much as a data pipeline problem
- — The best debugging tools answer user-experience questions, not only assertion status