Extract data from many PDFs into a table
Go to Extractions, choose a preset or define fields, pick documents and click Extract. Review each value against the quote it came from, approve or correct it, then export XLSX, CSV or JSON. Each page read costs 1 credit.
Updated 2 min read Counts toward ai credits
On this page (3)
Step 1: Open Extractions
In the sidebar under AI & automation, click Extractions.
Step 2: Choose the fields
Under Field source, pick a Saved schema, or New fields and start from a preset: Invoice, Receipt, Bank statement summary or Contract key terms.
Step 3: Adjust fields
Click Add field, give it a label and a Type: Text, Number, Amount, Date, Yes / no, One of a list or Table. Add a description of what to look for. Tick Save as schema to reuse it.
Step 4: Pick documents
Choose From library or Upload PDFs. Uploads are saved to your library.
Step 5: Run it
Click Extract … documents. Progress shows per document; you can Stop the batch or Retry failed.
Review the results
Every value has a status. Click any value to see the page and quote it came from, and to correct it.
| Status | Meaning |
|---|---|
| Found | The quote was found on the cited page and supports the value. |
| Unverified | The AI answered, but its quote or value couldn't be matched to the page text. Check it. |
| Missing | Not found in the pages that were read. |
| Invalid | A value was found but doesn't fit the field's type, options or pattern. |
| Corrected | You changed the value. |
Turn on Only documents that need a look to filter. Approve or Reject each document. Exports use your corrections and leave out rejected documents.
Export
Under Export, click XLSX, CSV or JSON. CSV has one row per document; Table fields (like invoice line items) go to a separate CSV per table, so a CSV export with tables downloads as a .zip.
Limits
- Extraction needs a free account; guests see a sign-up prompt.
- Each extraction reads up to 10 pages per document on Free, 50 on Pro and 100 on Business. Pages beyond that are listed as not read (Read … of … pages).
- Scanned PDFs need OCR first. Run OCR PDF on them, then extract.
- Results stay private to your account.
Can I run the same schema automatically?
Yes. Saved schemas are reusable in workflows and by connected agents, not just on this page.
I edited a schema. Do old results change?
No. Changes apply to new extractions; past results keep the fields they were made with.
Tools in this article
Did this answer your question?
Related articles
- Ask across several documents and make briefs, slides or quizzesCreate a workspace in Workspaces, add documents, and use Ask across documents (2 credits per answer) or Make something from these documents to generate a brief, slide outline, quiz or audio brief. Every citation names the document, version and page.
- How AI credits workEvery AI action costs a fixed number of credits, shown on its button before you run it. Your plan includes credits every month; an AI credit pack adds 500 credits for $5 that never expire and are used after the monthly ones.
- How accurate AI answers are, and how to check themAI answers can be wrong or miss nuance, so every result links to the pages it came from and chat quotes are checked word for word against the page. Check numbers, dates and obligations against the cited page before you rely on them.
- Automate repeated steps with saved workflowsOpen Workflows, pick a recipe or Start empty, add steps, choose how it starts, then click Create workflow. Each step uses its usual allowance (OCR pages, AI credits), and your plan sets how many workflows you can save.