Every bank change programme eventually asks the same question: how much testing should be automated? The honest answer is that it depends on how often a test will run and how stable the thing it tests is.
Automation earns its cost on regression: the hundreds of checks that must pass on every release, every patch and every close-of-business run. Once scripted, they run overnight, the same way every time, and free testers for work that needs judgement.
Manual and exploratory testing remain essential where requirements are still moving, where the user experience matters, and where an experienced tester’s instinct for edge cases finds the defects a script never would.
In practice we recommend a layered approach: automate the stable core (payments, postings, interest and fees, reconciliation outputs), test new features manually until they settle, and keep a small exploratory budget for every release. For Temenos estates, partner tooling such as TARIS can cut regression effort substantially.
The result is faster releases with less risk, and an evidence trail your auditors and regulator can follow.