Open Extract at the top of your library. A workspace owner defines the fields you care about once, such as an invoice total, a due date or a counterparty, and you run them across a set of documents. You get a table back: one row per document, one column per field.
Every value shows the page and the sentence it came from. Select a value to see its source, so you can check a figure without reopening the file. A field the document does not state comes back empty with a reason rather than a plausible-looking guess, and a value Seben cannot trace back to the page is discarded rather than reported.
Before a run starts, you see how many documents and pages it covers. A run takes up to 50 documents and 500 pages, and reads at most 40 pages of text from each document.
If you correct a value, the original is kept beside your correction, so over time you can see where extraction is reliable for your documents and where it is not. Export CSV downloads the results, including the page each value came from.