Skip to main content
When you don’t need to build an entire document, three focused endpoints let you work with short snippets of text: humanize a machine-sounding draft, rewrite a passage to a custom instruction, or draft a fresh section from a prompt. All three return structured JSON — no file is produced. These are the same helpers the Overten web app uses inside its inline editor, exposed for your own integrations with one sk_live_* key.

​
Humanize

Rewrite text so it reads like a person wrote it — keeps meaning, drops AI tells.
Response

​
Rewrite

Rewrite text to your own instruction — tone, length, style, audience, etc.
Response
The prompt field carries the instruction; the text field is the passage to transform.

​
New content

Draft a fresh section from a prompt. Useful for filling in a missing section of an outline before you generate the full document.
Response
file_name is an optional hint — the model uses it to tune tone/length to the target document type (a .docx briefing reads differently from a .pptx slide). Leave it out if you don’t care.

​
Request limits

For longer bodies, chunk on paragraph or section boundaries and stitch results client-side.

​
Errors

Same shape as the rest of the API:
See Errors for the full list.
These endpoints run on a smaller, faster model than /excel/generate et al. — most calls return in under 3 seconds. No sync/async split; you always get the answer in the response.