ai_questions.regenerate_embeddings
writes
(Re)compute semantic embeddings for every FAQ question using the tenant's current detection provider. No-op when the mode is 'lexical' (returns embedded=0) or the provider's API key is absent from the server env (returns missingKey=true). Calls the embedding provider API. Owner-only.
Input parameters
Returns object.
Output fields
| Field | Type | Required | Notes |
|---|---|---|---|
embedded | integer | yes | |
mode | string | yes | |
missingKey | boolean | yes |
ai_questions.report_false_positive
writes
Mark an FAQ question as a false positive — increments its false_positive_count, which feeds the detection-accuracy analytics. Use when the classifier matched a question it shouldn't have. Mutating.
Input parameters
| Field | Type | Required | Notes |
|---|---|---|---|
id | string | yes |
Output fields
| Field | Type | Required | Notes |
|---|---|---|---|
ok | literal | yes | One of: true |