Client Context
A digital bank maintaining large web, mobile, and API regression suites relied on Git repositories, Azure DevOps, API services, mobile apps, and test management to support payments, beneficiary management, cards, loans, and profile services. The operating model handled thousands of automated tests per release across South Africa, using code changes, defect history, test tags, and execution results. Full-suite runs delayed feedback and consumed significant CI capacity. Manual test selection occasionally missed high-risk dependencies.
Challenges
- Full-suite runs delayed feedback and consumed significant CI capacity.
- Manual test selection occasionally missed high-risk dependencies.
- Test teams manually translated requirements for payments, beneficiary management, cards, loans, and profile services into test cases and automation tasks.
- Large regression packs consumed excessive execution time across Git repositories, Azure DevOps, API services, mobile apps, and test management, even for small changes.
- Frequent application updates created locator failures, flaky tests, and high maintenance overhead.
Solutions Implemented
- Implemented a regression-selection agent using change mapping, dependency analysis, and defect history.
- Added explainable selection reports and mandatory critical-flow coverage rules.
- Deployed requirement-aware agents to generate reviewable functional, negative, and boundary scenarios for payments, beneficiary management, cards, loans, and profile services.
- Implemented impact-aware regression selection using code changes, defects, dependencies, and historical execution data.
- Introduced self-healing execution that re-evaluated locators and separated product failures from automation failures.
- Connected API, data-generation, evidence, and reporting agents to the existing CI/CD and test-management stack.
Value Delivered
- Reduced the primary testing or operational effort by approximately [70%], subject to validation against approved engagement data.
- Reduced execution time without sacrificing risk coverage.
- Reduced script-maintenance effort while preserving tester review and control.
- Expanded negative, edge-case, and cross-layer coverage.
Impact Highlights
- Achieved an estimated [50%] improvement in cycle time, coverage, or processing consistency; replace with the approved client metric.
- Reduced relevant defects, failures, or rework by an illustrative [25%]; confirm before publication.
- Faster pull-request feedback and more predictable release gates.
- Created a reusable agent framework for future products and teams.
