Send result to:
Copying values out of PDFs by hand is the whole job: open a file, find the number, type it into a spreadsheet, repeat sixty times. Converters turn a PDF into a sheet full of everything on the page, and chat tools hand back an answer you cannot verify. This extractor does one thing: it reads every document — digital text natively, scans through OCR — pulls only the fields you asked for, and shows you where each one came from.
Upload up to 10 files and each document becomes one row of the same table, with its file name in the first column. Ten invoices from ten suppliers, laid out ten different ways, end up as ten comparable rows — ready to sort, filter or paste into your books.
Write "pull the invoice number, date, supplier and total" and the columns are built for you, filled from one of your documents as a preview. Rename them, change a type, add one, mark one required — all before anything is processed, so a wrong guess costs you nothing. No templates, no zones to draw.
Pages with no text layer are deskewed and recognised before extraction, so a folder of scanned or photographed documents behaves like a folder of digital ones. The language of each scan is detected automatically — English, German, French, Spanish, Chinese, Japanese, Arabic and more than 30 others.
Each cell carries the document, the page and the exact spot it came from. Click it and the original page opens with the value highlighted. No summary, no paraphrase — the text that is actually on the page.
Low-confidence values are flagged instead of being presented as fact, and when a document states the same value twice with different contents, you see both. You check the handful of cells that need it rather than re-reading every document.
Any value a document states once: invoice number, date, supplier and total from invoices; store, date and amount from receipts; parties, signing date, term and amount from contracts; order numbers from purchase orders and delivery notes; name, email and phone from CVs and application forms.
Excel can import tables that already exist in a digital PDF, but it cannot read a scan, and it will not find a supplier name or a due date in running text. Here you name the fields and the AI finds them wherever they sit on the page, in every file of the batch.
Text extraction, recognition and the spreadsheet itself run on the same PDF engine as our other tools. Files that cannot be read are listed with a reason rather than dropped in silence.
The columns are agreed before anything is processed; then every document is read, scans through OCR, and the values land in one table you can check against the pages.
Files are deleted from our servers after 24 hours. The table and your edits live in this tab only.
Yes. It works in your browser with nothing to install, and processes up to 10 files and 200 pages per run.
Yes. You name the fields in plain words, the AI reads each document — scans through OCR — and returns the values as columns of a table. Each value is tied to the place on the page it was found, so you can check it in one click.
Upload up to 10 PDFs, describe the fields once and agree on the columns. Every document becomes one row of the same table, and the Excel file you download holds them all.
Yes. Pages without a text layer are straightened and recognised before extraction, so scanned invoices or photographed receipts saved as PDF work like digital files.
Digital PDFs in any language. Scans are recognised in more than 30 languages, including English, German, French, Spanish, Italian, Portuguese, Polish, Russian, Ukrainian, Turkish, Arabic, Hebrew, Hindi, Chinese, Japanese and Korean — the language is detected automatically, so there is nothing to choose.
Any value a document states once: an invoice number, a date, a supplier, a total, contract parties, a term, a name or an email. Each document gives one row. For long tables such as invoice line items or bank transactions, a PDF to Excel converter is the better fit.
Up to 10 PDF files per run, 10 MB each and 200 pages in total.
Click any cell to see the exact place on the original page it was taken from. Uncertain values are flagged, and you can go through them one by one with the N key.
Excel (XLSX), CSV or JSON. Dates stay dates and amounts stay numbers in Excel, and your corrections are included in the download.
Uploaded files are deleted from our servers after 24 hours. The table and your edits stay in your browser tab and are not stored.
