Skip to main content

Known limitations

We aim to make AIVA as widely useful as possible. Still, there are some not-yet-supported scenarios and apps that, unfortunately, aren't yet testable with AIVA because of missing features.

These are the main ones we're aware of:

LimitationPlanWorkaround
Using AIVA in Safaribased on feedbackUse a Chromium-based browser
Apps that require WebGLbased on feedback
Test scenarios relying on right click, drag-and-drop, zoom or other mouse/finger gesturesbased on feedback
Scrolling within small UI elements (e.g., combo boxes, drop-down menus or small modals) may not be reliable.Improve scrolling reliability based on user feedbackDuring recording, making the entire list visible and positioning the target item near the middle of the list may improve scrolling reliability. However, this approach may still fail in some cases.
Localization of elements within tables and grids can be unreliable in some scenarios.Improve localization strategy (In progress)See Reliable table interactions for recommendations on recording reliable table interactions.
By content requires a row that can be distinguished from other rows. If the row contains only repeated values (for example, the same status appears in multiple rows) or contains only numbers, prices, icons or images, AIVA may not be able to identify the correct row.Improve localization strategy (In progress)Record the step on a row that has distinctive text (e.g. name, description, email address), or switch the step to By position, if the row position is stable.
By position requires AIVA to detect either a table header or a table title. If neither is detected, By position is not available for that table.Improve localization strategy (In progress)Record the step with the table header or title visible. If that is not possible, use By content.
By position may fail if the table header or title used to identify the table is not visible both during recording and during replay.Improve localization strategy (In progress)Keep the table header visible together with the target row. For tables without a header, keep the table title and the top of the table visible.
Tables whose rows contain only icons or images, with no readable text, cannot be matched using By content.Improve localization strategy (In progress)Use By position, or record the step on a cell that contains readable text.
A module can’t contain another module (no nesting yet).Support nested modules (coming soon)Keep shared steps in a single flat module and reference it from tests.

For guidance on working with tables, choosing a Table search mode, and recording reliable table interactions, see Reliable table interactions.

We will be prioritizing removing these limitations the same way as any other development work: based on user feedback and actual use cases, so please do get back to us if any of them are blockers for you.