Log Front snippet gaps to Airtable when nothing fits

Every time a new customer message arrives in Front, this workflow checks it against your existing snippet library. When the best match still falls below the confidence threshold, it treats that as a content gap and records the question, inbox, and conversation link in an Airtable table dedicated to missing canned replies. Over time you get a ranked backlog of the questions your library cannot answer.

Use it when you already run a snippet suggester and want to close the loop on coverage. It turns the moments agents had to free-hand a reply into a prioritized to-do list for whoever owns your help content.

1. 1A new inbound message in Front triggers the run.
2. 2The message is matched against the snippet catalog and scored by the model.
3. 3A branch checks whether the top score clears the gap threshold.
4. 4If it does, the run ends quietly with no action.
5. 5If it does not, the question and metadata are written as a new Airtable row, deduped against existing open gaps so repeated questions increment a count instead of creating duplicates.