Q12 of 21 · AI for testing
How do you use AI to convert manual test cases into automated scripts?
Short answer
Short answer: Feed the manual test case steps to a language model alongside the target framework, existing Page Objects, and locator conventions. The model handles boilerplate conversion; you fix the selectors (it cannot look at the live DOM) and verify the assertions match real expected values.
Detail
Manual test cases often read like: "1. Navigate to /login. 2. Enter 'user@example.com' in the email field. 3. Click Login. 4. Verify the dashboard is shown." This is procedural enough for a model to translate to Playwright code reliably.
What the model cannot do: find the actual ARIA labels or data-testid values used in your application, or know what "dashboard is shown" means in concrete DOM terms. You have to provide that context — your Page Object classes, the selectors in use — or correct those parts yourself after conversion.
For large backlogs: batch-convert by functional area. Paste 5–10 related test cases at once with the same Page Object as context. Prioritise converting the highest-risk manual tests first — AI conversion gives diminishing returns on edge cases that require domain understanding the model doesn't have. See Requirements to test cases.