Self-Healing Test Automation Case Study

Customer: Australian bank
Industry: Banking and finance
Project: Core banking platform upgrade

The challenge

An upgrade that
could break everything.

Core banking upgrades change the screens and objects that automated tests depend on. With 1,383 screen changes in this upgrade, traditional coded automation would have broken on a massive scale, forcing the bank to rework scripts before it could regression test the upgrade.

Script rework on that scale takes specialist automation engineers weeks, delaying the upgrade or forcing the bank to fall back on slower manual regression at exactly the moment it most needed fast, reliable feedback on its core banking processes and customer channels.

The bank needed its regression automation to keep working through the upgrade, so it could validate the new platform quickly and move to production with confidence. Any delay would hold up the upgrade and keep the bank on its older platform for longer.

Analysts comparing core banking screens side by side

What made it hard

1,383 changed screens: The upgrade changed 1,383 screens that the automated tests interacted with, a scale that would normally break most scripts.
Coded scripts break easily: Traditional automation is coded to each screen, so it fails when screens change and needs weeks of script rework.
Regression needed immediately: The bank needed full regression results quickly to validate the upgrade before it reached customers and branches.

Our solution

Machine learning instead
of script rework.

The bank's automation ran on Enginuity, whose natural-language scripts are not coded to individual screens. Enginuity's AI scanned the upgraded application, detected the changes and relearned them in 2 hours, so the existing tests continued to run.

Enginuity's machine learning updated the AI automation for all 1,383 screen changes in 2 hours. Because tests describe business steps rather than technical objects, the scripts themselves did not need changing, and no automation engineer had to rework the test pack.

With the automation updated, the full regression pack could run against the upgraded platform straight away, giving the bank rapid evidence that its banking processes still worked. The upgrade could proceed on schedule, supported by complete regression results rather than partial manual checks.

How it was delivered

From upgrade to
regression in hours.

Enginuity's self-healing kept the regression pack usable through the upgrade.

01
Design
- Upgrade scanned
  Enginuity's AI scanned the upgraded core banking application to detect every changed screen and element before regression began.
- Changes relearned
  Machine learning relearned the 1,383 changed screens in 2 hours, with no manual script maintenance by the bank's team.
02
Execution
- Scripts reused
  Existing natural-language scripts ran unchanged against the new platform, so regression started straight away after the upgrade.
03
Verification
- Regression completed
  The full regression pack validated the upgrade without any test pack changes, giving the bank fast evidence to release.