The two highest-stakes operations on a relationship database are bulk import and bulk action.
A CSV with the wrong column mapping can write a thousand bad records in a second. A rhythm analysis that mis-classifies which relationships need attention can put a hundred people into a follow-up queue they should not be in. Both feel like single-click actions. Both can take an afternoon to undo.
Founding users were doing both through Assist within the first week and asking, reasonably, “wait, what is this actually going to do?” The previous flow ran end-to-end and showed you the result. The result was usually right. Sometimes wrong. There was no point to intervene.
Assist now treats bulk operations as guided workpieces.
A CSV import is a named-step workflow. Parse, map, dedupe, enrich, score, save. You confirm each step before it runs. You can scope the import to a subset of rows (say, the first 10) to test the mapping, then re-run on the full file. Nothing writes to your workspace until the final confirmation. If a column maps to the wrong field, you find out before you’ve written it to a thousand records.
A Relationship Rhythm analysis is the same shape. Assist proposes which relationships need attention based on the 8.33x Rhythm-Break Rule. The proposal lists each relationship, the evidence (last contact, last meaningful exchange, decay risk), and the recommended action. You approve in batches or individually. Nothing enters your queue until you confirm.
The save action is always the final, explicit step. There are no implicit writes.
What changes for you. Bulk operations stop being scary. The “what is this about to do” question gets answered before the operation runs, not after. The “wait, let me undo this” version of an afternoon goes away.
