Synchronously calls the ReviewData AI provider to draft a reply for a review, persists an AiReviewResponse row, and counts 1 unit against the store's ai_response plan quota (hard-stop plans return 403 once exhausted; soft-stop plans bill the call as overage).

Auth: Both — authentication. The target store_id/review_id must sit inside your own account tree: the store (and the review's store) must resolve to a company that is the authenticated user's own person.company_id or a descendant of it. Partner (bundle_id=1) and Account (bundle_id=3) roles use the same endpoint; only the reachable part of the account tree differs.

Rate limit: No rate limit.

Request

• Body:

The controller also pulls review_text and rating from the resolved review itself — they are not client-supplied.

Response — { data. } envelope. On success the controller returns { success: true, data: <service result> } wrapped, plus ai_response feature headers via withFeatureHeaders.

• data.success (boolean) — always true on the happy path.
• data.data.response_text (string) — the generated reply text.
• data.data.model (string) — model id used, e.g. gpt-4.1-mini.
• data.data.tokens (object) — { input, output, total } token counts (integers).
• data.data.task_id (string) — ReviewData task id.
• data.data.foreign_key (string) — RD foreign key.
• data.data.ai_review_response_id (integer) — id of the persisted AiReviewResponse row.
• data.data.remaining (integer|null) — remaining quota, null when uncapped.
• data.data.overage (boolean) — whether this call was billed as overage.
• data.data.usage_log_id (integer) — usage record id.

Failures: 403 quota exceeded / feature disabled, 422 validation, 502 provider unreachable; review-not-found returns an error envelope.

Errors: 422 on validation failure.