Maintaining data integrity with pipelines and grid edit

I have a pipeline where the first step searches for a record in a parent table. If no match is found, it creates the parent record. If a match is found, it skips creating the parent record and moves on to add the child record. This works great when used in Edit mode from a single record, but causes problems when someone tries to enter multiple records using Grid Edit mode. Because the pipelines run virtually simultaneously, I have seen cases where duplicate parent records are created. I can mitigate this by adding a unique field to detect duplicate records, but when this occurs the pipeline for that record will stop and the child record will not be created.

What is the magic method to ensure data integrity when adding records through pipelines while using Grid Edit mode?

------------------------------
Brian Petzold
------------------------------