← Back

Behind the Scenes

Same problems, two eras. How work was done at Newfold Digital (traditional SDET) vs how it's done now with agentic AI workflows.

Traditional vs Agentic: Side by Side

TaskTraditional (Newfold)Agentic (Current)
Write test for new featureManual Cucumber feature file + step definitions + page objects (~2-4h)AI generates from Acceptance Criteria → human reviews + adjusts (~20min)
CI/CD pipeline configJenkinsfile hand-coded with Groovy syntax, tested via trial-and-errorAgent proposes pipeline → human approves + adjusts parameters
Framework architectureResearched patterns, POC'd, coded layers over weeksKATA framework scaffolded in hours via skills, composable from day one
Test reportingCustom reporter code per project, HTML/XML output formatsKATA reporter + Allure out of the box, historical trends baked in
Git workflowManual branching, conflict resolution, PR managementgit-flow-master skill handles branching + commits + PRs autonomously

Architecture Evolution

Two pipelines — same engineer, expanded toolkit.

Architecture diagram comparing Traditional SDET and Agentic QA pipelines

2019-2023: Traditional SDET

  • •Cucumber + TypeScript Page Object Model
  • •Robot Framework + Python for API tests
  • •Jenkins pipeline with manual stage config
  • •Allure reports, Jira traceability
  • •Manual Cucumber feature file authoring

2025+: Agentic QA

  • •KATA framework with component-action-test layers
  • •AI-driven test generation from AC Gherkin
  • •GitHub Actions CI with multi-stage orchestration
  • •Autonomous mode gate bypass for sprint workflows
  • •Health tooling: asset integrity, staleness, consistency

This Portfolio — Built with Agentic Workflows

Everything you see on nelthor.qzz.io was planned, built, reviewed, and deployed using the agentic-dev workflow. The process:

  1. Design spec defined via/project-foundation
  2. Content strategy via/product-management
  3. Landing page scaffolded via/project-bootstrap
  4. Each feature shipped via/sprint-development
  5. Quality gates via health tools +bun run health:check
  6. Autonomous mode enabled for faster iteration