Extract data from PDFs
Contracts that need logging, invoices that need totalling, hundreds of notices where you need one clause — the content sits in PDFs, but analysis needs a table.
How you probably do it today
Open a PDF, find the amount, copy, switch to Excel, paste… Two minutes per file, three hours for a hundred, and rows get shifted along the way.
Just say this
Read the PDFs in the "Contracts" folder and extract the client, supplier, amount and signing date into one Excel sheet.
Or more specifically:
Extract invoice number, issue date, amount and tax from these invoice PDFs, sorted by issue date.
What it will do
- Try one file first — shows you the fields and the values it found, so you can confirm
- Then process the batch with the same set of fields
- Flag what's uncertain — anything it can't read is left blank and the file is named
What you get
- One table, one row per file
- The file name in the first column, so you can always trace back to the original
- Missing values are left empty and listed separately — it will not invent a number
- A warning about which files were scans and may be less reliable
Other things to ask for
- "Quote the exact paragraphs that mention 'penalty' in each contract"
- "Group by company and total the contract value for each"
- "List the ones where the amount and the date don't match"
Digital PDFs vs. scans
A digital PDF (exported from Word or a system) already contains text, so extraction is accurate. An image PDF (photographed or photocopied) needs text recognition first; handwriting and text under a stamp will contain errors. Start with a small batch and spot-check a few before running all.
What it cannot do
- Handwriting and text hidden by stamps will be misread
- Multi-column layouts and nested tables may need a few rows checked by hand
- Encrypted PDFs must be unlocked first
- It won't make judgement calls — spot-check important files yourself