Use AI” Action Generates Empty NaN Rows and Consumes Credits When Run on Empty Inputs

Description:
When running the “Use AI” action on all rows of a table that contain empty or null values, the system generates dozens of NaN rows instead of skipping them or returning a controlled empty output. This results in unnecessary consumption of AI action credits without producing any usable data.

Steps to Reproduce:

1. Select a table where the input columns for the AI action contain empty rows.

2. Configure the “Use AI” action to run on all rows.

3. Execute the action.

4. Observe that multiple NaN rows are generated in the output, even for rows with no input data.

Expected Result:

• Empty rows should be automatically skipped or ignored,

• or return a controlled empty result ([] or "") without consuming AI credits.

Actual Result:

• NaN rows are generated for each empty input row.

• AI credits are consumed for these empty rows.

• No useful data is returned from these executions.

Impact:

• Unnecessary AI credit consumption.

• Table is filled with NaN values, which may cause confusion in downstream workflows or exports.

• Running the action at scale is inefficient and wasteful.