TestMu AI has introduced a new KaneAI test-authoring engine that represents tests as objectives rather than fixed click paths. When an interface changes, it can re-author affected steps, but reviewers can inspect the result before it replaces the existing test.
Watch Desk analysis
What happened
The new engine, called New Experience, is available in KaneAI and currently supports desktop web browsers. TestMu says it underpins two options: Adaptive Heal re-authors the objective containing a failed step and those after it, while Dynamic Test builds every objective from its intent on each run instead of replaying recorded steps.
Re-authored versions appear as drafts in version history. A reviewer can approve or decline them, and approving regenerates the exported code. Automatic approval is switched off by default. The release announcement also says the features require Chrome and tests authored with New Experience; other test cases continue to replay their recorded steps. Re-authoring consumes authoring credits.
The announcement in The Star describes the change as a way to keep tests useful when an interface is restructured, not merely when a selector is renamed.
Why it matters
A test tied to a particular sequence of clicks can break when a product’s interface moves on. Rebuilding around the test’s objective could make maintenance less brittle, while the draft-and-review step gives teams a chance to catch a rewritten test that might otherwise hide a real failure.
There are boundaries: the new behaviour is not described as applying to every test or browser, and teams will need to account for authoring credits. This is a meaningful change to how UI tests can be maintained, not a promise that testing has become self-running and worry-free.
Our read
The interesting choice is to let an AI rewrite the route while keeping a person responsible for accepting the destination. That is a better starting point than treating a self-healing test as trustworthy simply because it passes. Teams using KaneAI should check whether their tests qualify, and keep the approval gate in place until they have confidence in the results.
What to watch
- Whether New Experience expands beyond desktop web and Chrome.
- How often re-authored tests preserve the original test’s intent in real applications.
- How authoring-credit use affects the cost of maintaining larger test suites.
Discussion spark: Should AI-rewritten tests always require human approval, or is there a point where teams should let them replace the old version automatically?
Sources and evidence
- Source update (19 October 2026, 14:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.