POST a URL. Get a structured transcript back.
One endpoint. Audio, YouTube, or a Stage Mode session ID. Same response shape every time.
The whole product, as endpoints.
Every surface that exists in the ActaScribe app is also a REST endpoint. No private APIs.
POST /v1/transcriptsCreate a transcript from audio, a YouTube URL, or a Stage Mode session. GET / DELETE supported for full CRUD.
POST /v1/templatesDefine a custom 8-section template, override system templates, or list the ones your workspace already has.
POST /v1/talkbase/corporaCreate a TalkBase corpus. Add sources, query the corpus, manage embeddings, set retention.
POST /v1/stage/sessionsSpin up a Stage Mode session programmatically. Get the public URL, the WebSocket handle, and a CSV of audience Q&A.
POST /v1/webhooksSubscribe to events: transcript.completed, corpus.updated, stage.session.ended. HMAC-signed deliveries.
Typed clients for the languages you ship in.
First-class TypeScript and Python clients. More on the way.
$ npm install @actascribe/sdk$ pip install actascribeBoth SDKs publish around general availability — installation copy shown here is the future-state target. The REST API is live today.
Predictable, signed, retryable.
What grown-up webhook infrastructure should look like in 2026.
HMAC-signed
Every delivery is signed with your workspace secret. Verify the X-ActaScribe-Signature header before processing.
RFC 7807 errors
Standard problem+json error envelopes for every failure. Type, title, detail, instance — same shape across every endpoint.
Idempotency keys
Pass an Idempotency-Key header on every POST. Retries are safe; duplicates return the original response.
Quotas you can see in the headers.
100 requests/min for standard endpoints. 10 requests/min for heavy endpoints (transcript creation, Stage Mode session spin-up). Every response includes X-RateLimit-Remaining and X-RateLimit-Reset so you can pace your client without guessing.
ActaScribe,
as an API.
Included on every paid tier. No separate API plan to negotiate.
