Partner apps built on the Sure Send API can now transcribe a recording. Point POST /api/partner/ai/transcribe at an audio or video file and you get back a transcript split into utterances, each one labelled with who was speaking and timed in seconds, so a tool can show time-coded lines that jump to that moment in the recording.
The file is fetched straight from wherever it already lives: you send an https link, such as a presigned S3 or Cloudflare R2 URL, and nothing has to be uploaded to Sure Send first. Video files work as they are, with no need to extract the audio.
Transcription is billed to the same AI Credits balance as AI grading, priced per whole minute of audio. GET /api/partner/ai/credits?objectType=transcription&durationSeconds=… tells you what a recording will cost before you start it, so a partner app can show the price up front. Credits are reserved when the job starts and settled against the length the recording actually turns out to be, so you are charged for the audio that was transcribed and never more than you were quoted. A job that fails, or that never finishes, costs nothing.
Long recordings run in the background: starting a job returns immediately, and GET /api/partner/ai/transcribe/{id} reports processing, completed, or failed along with the finished transcript.
Widget and OAuth tokens need the use_ai scope, which now reads Run AI grading and transcription using team AI credits to reflect what it covers.