They can describe the final accepted condition and the recorded process route, but not the unmeasured condition before correction. Establish the drawing revisions and measurement methods before using any in-jig check as a comparison with the final figures.
Also check whether the jig card follows identifiable parts or only the batch, since linking a clamped observation to an individual final result requires more than finding them in the same folder.
Is correction labour still in the proposed route? I would keep an allowance while the review is unresolved, rather than assume the robot fixture makes that work disappear.
That helps distinguish the stages. Have fabrication and inspection agreed a current-part trial that includes the first released reading before anyone corrects it?
Include incoming component fit-up in that trial as well, with the actual current revisions, so the fixture designer can assess what the nest is being asked to locate rather than only the shape after welding.