Send an agent to extract data from a document, optionally transforming it to a specific data schema. POST a file called ‘file’. Must be a filetype readable from the dashboard document task (ie pdf or image)
Bearer authentication header of the form Bearer <token>
, where <token>
is your auth token.
Additional hints to the AI about what you want to extract
The model to use. Large is slower but may be more accurate. Small is faster but may be less accurate.
auto
, large
, small
Attempt to extract extended metadata, like URLs and images
Increasing this will cause the AI to scroll further down the page
A JSON schema to extract against
{ "schema": { "name": "<the site name>" } }
An in-progress result, which you can retrieve later with /result/{result_id}
The response is of type object
.