Bindery

100% local

How to Extract Text from a PDF for ChatGPT & Claude

You have a PDF — a contract, a research paper, a manual — and you want ChatGPT or Claude to summarize it, answer questions about it, or translate it. The missing step is getting clean text out of the PDF. Here is how to do it in about a minute, without uploading your document to anyone's server.

The short answer

Open Bindery's free PDF → Text tool, drop in your PDF, and copy the extracted text. Then paste it into ChatGPT or Claude with a clear instruction, for example: "Summarize the key points of the following document." Extraction takes seconds, and your file never leaves your browser.

Step by step

  1. Open the PDF → Text tab.
  2. Drag your PDF onto the drop area, or click it to browse for the file.
  3. The text appears automatically — page by page, in reading order. Note the character count shown; it matters for long documents (see below).
  4. Click "Copy text" to copy everything, or download it as a .txt file to keep.
  5. In ChatGPT or Claude, write your instruction first, then paste the text underneath it.

Why instruction first? Models follow leading instructions more reliably than instructions buried after a wall of pasted text.

Long PDFs and context limits

AI chats can only absorb so much text at once. Paste a 200-page report in one go and part of it may be ignored or summarized unevenly. Work in sections instead:

  • Split at natural boundaries — chapters, headings, or roughly every 10–20 pages.
  • Ask the same question of each section, then combine: "Combine these section summaries into a single overview."
  • If answers start mentioning things that are not in the document, the paste was probably too long — shorten it and try again.

Prompt templates that work

Copy, adapt, and paste one of these before your extracted text:

  • Summarize: "Summarize the following document in five bullet points, focusing on decisions, dates, and obligations:"
  • Ask questions: "Based only on the document below, answer this question: [your question]. If the answer is not in the document, say so."
  • Translate: "Translate the following document into [language], preserving headings and list structure:"
  • Pull out data: "Extract every [name / date / amount] mentioned in the document below and list them in a table:"

The "based only on the document" clause is the most valuable phrase on this page — it cuts down invented answers.

Clean up messy extraction

Extracted text is rarely perfect. Two quick fixes cover most of it:

  • Repeated headers and footers appear on every page. Delete them, or ask the AI to do it: "Clean up this extracted text: remove repeated headers and page numbers, don't change anything else."
  • Hyphenated line breaks ("docu- ment") confuse models slightly. Joining them back into whole words takes a minute with find-and-replace.

When nothing comes out: scanned PDFs

If the tool reports no extractable text, your PDF is probably scanned images rather than real text — common with older documents and phone scans. Text extraction cannot read images, and no setting will change that. Your options: run it through OCR software (many PDF readers and operating systems include it), or re-export from the original source. Bindery tells you plainly when there is nothing to extract instead of handing you garbage.

A privacy warning worth heeding

Most free "PDF to text" websites upload your file to their servers to process it. For a public manual that is harmless. For a contract, medical record, or financial statement, it means your sensitive document now sits on someone else's computer, outside your control. If the document is sensitive, extract it locally — in your browser with a tool like Bindery's, or with software installed on your machine.

FAQ

Can ChatGPT read PDFs directly?
Paid plans allow file uploads with mixed results. Pasting extracted text works on every plan, including free, and gives you control over exactly what the model sees.

Will tables and formatting survive?
No — you get plain text. Tables become linear text, which is usually fine for summarizing but awkward for precise data work.

Is there a file size limit?
Processing happens in your browser's memory, so very large PDFs (hundreds of megabytes) may be slow. Ordinary documents are instant.

Does this work on my phone?
Yes — the tool runs in the mobile browser the same way.

Try it now

Open the PDF → Text tool — free, no signup, your file never leaves your device. For the other tools, see the how-to guide.