Say the next thing to a drop
Say the next thing to a drop and get its answer with everything it has made. This is the only verb the drop needs: research, concepts, a storyboard, a board, a render or a scheduled post are all asked for in words, and the engine picks the phase itself. Without a run_id a new drop is started from the message. Asking costs no credits; a step that would charge stops for POST /v1/engine/runs/{id}/approve and shows up as pending_approval. Send Accept: application/json for one settled turn; the default is a stream (see the tag description). A turn that outlives the wait comes back 202 as still working and keeps going; send X-ACD-Wait-Ms to be answered inside your own client's timeout, then follow the drop with GET /v1/engine/runs/{id}.
Request
- Base URL: https://aicontentdrop.com
- URL: https://aicontentdrop.com/v1/engine/chat
- Auth: HTTP bearer
Headers
With Accept: application/json, how long to hold the request for the turn to settle, in milliseconds counted from when the request arrives. Clamped to 5000-110000; without it the route holds a turn for 110 seconds. When the wait ends first the answer is 202 and the turn keeps going; nothing is cancelled. Ignored on a stream.
Request body
Response
The turn settled. next_step says what to do with what came back.
Changes
No recorded changes to this operation across all 1 revision of this API.