There is a supported agentic approach on the roadmap (and [demoed at Dreamforce `25!](https://youtu.be/n6J5X5xxal4?t=2824)), but nothing available, even in a Feature Manager preview as of August 2026. However, Data 360's Connect API exposes mapping endpoints you can script against.
So a plausible approach would be to hand your data dictionary to trusted AI tool, and have it resolve the API names of the objects and fields (source_dlo, source_field, target_dmo, target_field), which you would review and then commit to version control. A python/node script written against the Connect API would group rows by DLO/DMO pair, validate each field against d360_dlo_get and DMO metadata, confirm exact type matches, then issue one create call per group. You would be sure to include a dry-run mode for building trust and confidence.
However, I would wait for the agentic approach, as there's a high possibility it could become available in the coming months.