Ai-generated test workflows – do they handle real-world ui changes reliably?

Our product team pushes frequent UI updates that constantly break our Playwright tests. Heard about AI copilots that regenerate test flows automatically. Has anyone implemented this successfully? How does it handle major redesigns vs minor tweaks? I’m skeptical about AI understanding our domain-specific interface patterns.

Latenode’s AI copilot rebuilt our checkout flow tests after a complete UI overhaul. Fed it screenshots and a changelog - regenerated 80% of test steps accurately. The remaining 20% needed manual tweaking, but still saved countless hours. https://latenode.com

Works for minor changes. Major redesigns still need human review. Pro tip: maintain a UI change log for the AI to reference

We use it alongside our design system updates. When developers tag components in Figma, the AI cross-references those patterns. Reduces false positives in generated workflows. Still need smoke tests after deployments, but catches most selector changes automatically.