Back to case studies
AI-Assisted Test Design and Review for a Fleet Telematics Platform
Client name withheld · Transportation
Unit tests
426
In production
v1.3, used by QA
Design time (target)
2–4 hours
15–30 minutes
For a fleet telematics platform with 37+ modules, 5 environments and 2 regions: a knowledge layer, not a test suite, that turns a Jira story into reviewed, scored test cases and uploads them to test management with a full audit trail.
What changed
Before
Test design
By hand
2 to 4 hours
After
Design
Design skill
10 coverage lenses
Review
Review skill
scored 1–5, gate at 4.0
Gate
Coverage gate
every criterion covered
Upload
Test management
dry run · dedup · no deletes
A Jira story goes in; design, review and upload skills turn it into scored test cases traced to its acceptance criteria, and a coverage gate stops anything uncovered.
What we built
The framework turns a Jira story into reviewed, scored test cases and uploads them to test management. It runs in Claude Code and GitHub Copilot from one source of truth, and every case lists the acceptance criteria it covers.
Key Features
Six skills, each with a Copilot agent twin: design, review, upload, case 360 report, live preview and a design-then-review chain.
Two-part generation: an analysis (10 coverage lenses, scenario and risk matrices, a requirement inventory and a conflict table), then test-management YAML.
A deterministic coverage gate fails when an acceptance criterion has no case or a conflict has no human decision.
Review scores every case from 1 to 5; the gate needs an average of at least 4.0 and no blocker.
Upload confirms each section, shows a dry run first, deduplicates by ID or title, and never deletes.
An append-only audit trail: a conversation log, a transcript of every tool call, and an offline HTML viewer.
Results
Version 1.3 in production use by QA, after 104 commits in 5 months.
426 unit tests behind the framework, distributed as two npm packages with an install verifier.
Target: test design from 2 to 4 hours down to 15 to 30 minutes. This is a goal, not yet a measurement.
Stack
Engagement: 5 months · AI-Assisted Engineering
Next step
Let’s look at your SDLC together.
45 minutes with a senior engineer: your setup, the risks, what to fix first. No sales team.