Creative API

Premium GenAI creative generation backed by module-f (Vertex Imagen / Veo) and the fleet orchestration chain module-a → module-BDE → module-f. Routes are served under /v1/creative.

Async operations return a job_id; poll it at GET /v1/jobs/{job_id}.

Authentication

Authorization: Bearer ai92_live_<key>     # production
Authorization: Bearer ai92_test_<key>     # sandbox

Mutating requests support Idempotency-Key and are brand-gated.

Credit costs at a glance

POST /v1/creative/generate

Cost: Image = 50 credits. Video = max(400, 50 × duration_s) — i.e. 8s/12s/16s/20s → 400 / 600 / 800 / 1000 credits.

Generate creative assets. Image runs the premium DCO pipeline; video runs Veo. Credits are deducted after a successful render (no charge on backend failure).

Sandbox timing. Creative generation is asynchronous (the call is accepted and returns a job_id prefixed f_; poll it at GET /v1/jobs/{job_id}). In the sandbox the render pipeline can be slow and a job may reach status: TIMED_OUT. When that happens the reserved credits are automatically refunded and you can simply retry the call. Production turnaround is significantly faster.

Auth: Bearer

Request

curl https://api.ai92.ai/v1/creative/generate \
  -H "Authorization: Bearer ai92_test_51a9c0de…" \
  -H "Content-Type: application/json" \
  -d '{"brand_id": "acme", "prompt": "cinematic product hero shot, golden hour", "asset_type": "video", "duration_s": 12}'

Response 200 OK

{
  "asset_id": "ast_v12",
  "type": "video",
  "duration_s": 12,
  "url": "https://cdn.ai92.ai/ast_v12.mp4",
  "credits_charged": 600
}

Errors

GET /v1/creative/assets

Cost: Free

List generated creative assets.

Auth: Bearer

_No parameters._

Request

curl https://api.ai92.ai/v1/creative/assets \
  -H "Authorization: Bearer ai92_test_51a9c0de…"

Response 200 OK

{
  "assets": [
    { "asset_id": "ast_v12", "type": "video", "url": "https://cdn.ai92.ai/ast_v12.mp4", "created_at": "2026-06-20T10:00:00Z" }
  ]
}

POST /v1/creative/newsletter

Cost: 15 credits (async)

Kick off async newsletter banner generation. Credits are reserved now, consumed on success, refunded on backend failure. Poll the result at GET /v1/jobs/{job_id}.

Auth: Bearer

Request

curl https://api.ai92.ai/v1/creative/newsletter \
  -H "Authorization: Bearer ai92_test_51a9c0de…" \
  -H "Content-Type: application/json" \
  -d '{"brand_id": "acme", "topic": "Summer product roundup"}'

Response 200 OK

{
  "job_id": "f_7c1d0a92…",
  "status": "queued",
  "service": "creative",
  "credits_reserved": 15
}

Errors

POST /v1/creative/campaigns

Cost: 250 credits per 7-day week (async) — 250 × ceil(window_days / 7); planning window capped at 31 days

Generate a full creative campaign (a calendar of premium assets). 923 orchestrates the fleet chain module-a → module-BDE → module-f and returns a job to poll at GET /v1/jobs/{job_id}. Credits are reserved now, consumed on success, refunded on any failure.

Auth: Bearer

Request

curl https://api.ai92.ai/v1/creative/campaigns \
  -H "Authorization: Bearer ai92_test_51a9c0de…" \
  -H "Content-Type: application/json" \
  -d '{"brand_id": "acme", "planning_window_start": "2026-06-20", "planning_window_end": "2026-07-03", "schedule_type": "MOD80", "channels": ["instagram", "email"]}'

Response 200 OK

{
  "job_id": "bde_4af09c11…",
  "status": "queued",
  "service": "creative",
  "credits_reserved": 500
}

Errors