Home · Business · Admin & Office · Documents & templates

YES

As of 13 August 2026, AI can extract data from a PDF.

This still needs a person who signs their name to it.

Can you do it?

5 minutesto a draft.

30 minutesto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ityou

What the alternative costsNo priced human alternative is provided in the available tool data.

If this goes wrong: a missed or altered value enters your business records and causes a later reporting, payment or operational error.

What to actually do

  1. Hand it to a person

    The route this page recommends

    A person who owns the outcome does this end to end, worth it when the failure is dear.

  2. Use a tool built for this

    Second choice
  3. Do it yourself

    The distant third

    A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

    How to actually do it

    1. Open the PDF in a viewer and note whether it contains selectable text, scanned pages, tables, or handwritten content.
    2. Decide the exact fields and one-row-per-record structure you need, then write them in the prompt's field list.
    3. Upload the PDF to a chatbot or a purpose-built PDF extraction tool and paste the complete prompt.
    4. Export or copy the returned table into the spreadsheet, database or document where the extracted data will be used.
    5. Open the original PDF beside the extracted table and compare every row, figure, date, reference and unit against its source page.
    6. Correct transcription errors, investigate every [UNCLEAR] marker, and check that the number of extracted rows matches the number of records in the PDF.
    7. Send or import the checked table only after a colleague or process owner has confirmed any values that affect payments, reporting or customer records.

    Prompt

    Extract the requested data from the attached PDF. I need these fields or columns: [LIST THE FIELDS]. Use these pages or sections if known: [PAGE NUMBERS OR SECTIONS]. Return the result as a clearly labelled table, with one row per record. Preserve the wording, spelling, punctuation, units, currency symbols and decimal places exactly as shown. Do not infer missing values, correct apparent mistakes or combine rows. Leave a cell blank when the PDF does not contain the value, and mark text that is unclear as [UNCLEAR] with the page number. After the table, list every field or row that needs manual checking, including any possible OCR errors, merged cells, cut-off text or ambiguous figures. Include the source page number for every row or value where practical. Do not add information from outside the PDF.

    Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

What it gets wrong

  • AI cannot reliably resolve blurred scans, handwriting, cropped text or tables whose visual layout carries meaning.
  • AI cannot know whether a number is plausible for your business or whether two similarly named fields should be combined.
  • AI can alter a digit, decimal place, date or currency symbol while presenting the result in a clean table.
  • AI cannot take responsibility for records, reports or decisions made from an incorrect extraction.

Even on a YES, the friction has a name: verification cost, stakes of error and context depth.

How we scored this

Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.

AxisScore (0–2)
Output2
Inputs2
Verification1
Liability1
Effort delta2
Total8 / 10

FAQ

Can ChatGPT extract data from a PDF?
Yes. It can extract text, fields and many tables when you upload the PDF and specify the required structure. Check the result against the original page by page before using it in business records.
Can AI extract tables from a scanned PDF?
Often, but scanned pages depend on the quality of the image and the clarity of the table. Expect to check figures, decimal places, merged cells and unclear text manually.
What is the best AI tool for extracting data from a PDF?
PDF.ai is a purpose-built option for asking questions and extracting information from PDF documents. A general chatbot can also work, but neither removes the need to compare the output with the source.
Is it safe to upload a business PDF to AI?
Only upload it if your organisation permits the service and the document contains no information that must stay within approved systems. Remove unnecessary personal or confidential data, and check the tool's handling terms before uploading.

Nearby answers

Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.

The newsletter

AI news, new answers and product picks, straight to your inbox.