Documentation
How to authenticate, extract a page into structured data, and read the record behind every figure. Start with the quickstart. Currently early access.
Start here
- Quickstart — a key, one request, a result
- Extraction schemas — describing what to pull off a page
- Errors and limits — what each failure means and what to do
The shape of it
One endpoint does the work. You give it a URL and a description of what you want; it fetches the page from an address in the country you asked for, runs three extraction tiers cheapest-first, and returns structured data with a record of how it got there.
POST /v1/extract
Authorization: Bearer wi_live_…Everything else — browser sessions, exports, stored artifacts — exists for cases the single call does not cover.
What comes back
Alongside your data, every response carries the tier that produced it, every tier that was attempted with its outcome, the HTTP status of the fetch, and a request identifier. A figure you cannot trace is a figure you cannot defend, so the trace is not optional and not a separate call.
Last updated