Brand Studio docs
The APIs

OpenAI

One language-model read of your own copy, turned into voice claims that each have to quote your text or be dropped — and, before any money is spent, a second opinion on whether a candidate business is really you.

The facts a reader needs

What it unlocksThe fallback door for four of the five model calls — the voice read, the confirm-step adjudicator, the quality review and the confirm-step suggestions — on a deployment that holds no Gateway key. The fifth, the AI-visibility grid, needs the Gateway and is parked in the alpha.
When it is calledGateway-routed voice, legacy quality review and confirm-step calls prefer the AI Gateway; OpenAI is their fallback when AI_GATEWAY_API_KEY is absent and OPENAI_API_KEY is present. Client-review AI fixes separately require OPENAI_API_KEY and explicit owner action, with token preflight and one durable attempt per request, regardless of gateway configuration.
Cost basis$0.002 per pull for voice (gpt-5-mini: $0.25/1M in, $2.00/1M out) · $0.033–0.049 per pull for the quality review (gpt-5, measured over three live runs 2026-08-18). Client-review suggestions have a separate $0.05 conditional per-call ceiling and recorded token-usage estimates; billed USD stays unknown.
EnvironmentOPENAI_API_KEY
Bring your own keyYes.

The brand adjudicator on the confirm step (POST /api/brand-check), which runs before a pull is charged and may only demote a match, never promote one — and the quality review, which reads every settled section back against its own evidence and may only flag. Owner-triggered client-review AI fixes use direct Responses with one retained item and client message; suggestions never apply automatically.

Endpoints we call

EndpointWhy we call it
POST v1/chat/completionsEvery model call in the product, when the Gateway key is absent. Strict JSON schema, token-capped.

Checks it powers

Their documentation · get a key

← Every API

On this page