Repeatability
High
Data normalization and deduplication follow deterministic rules that apply uniformly across all rows. The same logic — strip text from SF values, standardize formatting, flag or merge duplicates — runs identically on every record, making this highly automatable.
Ambiguity Tolerance
Medium
Most success criteria are crisp (consistent SF format, no duplicates, clean CSV), but 'incomplete tenant names' introduces judgment: should the agent leave blanks, flag them, or attempt to infer? Duplicate-matching rules also need a clear definition (exact vs. fuzzy match on address) before the agent can proceed confidently.
Data & Tool Availability
High
The data lives in a Google Sheet, which is easily accessible via the Sheets API or a direct export. A code agent can read, transform, and write back the data or produce a CSV without needing any external context beyond the sheet itself.
Error Cost
Medium
Incorrectly merging or dropping legitimate records could corrupt the broker's CRM sync, causing missed follow-ups or lost listings. However, the original sheet is preserved as a backup, so errors are reversible with a human review pass before the final import.
Human Judgment Required
Low
No relationship context, taste, or ethics are involved — this is pure data hygiene. The only human input needed is a brief upfront spec on how to handle ambiguous duplicates and missing tenant names, after which the agent can execute autonomously.