Guide · AI
How to ask questions across multiple PDFs with citations
Reviewed
Short answer
To ask questions across multiple PDFs, collect them in a PDFBetter document workspace, then type your question under “Ask across documents”. The answer draws on every included document and cites the document, version and page for each claim, with a quote checked against that page; if the documents don't cover the question, it says so instead of guessing. Each answer costs 2 AI credits.
On this page (6)
Step by step
- 1
Create a workspace
Open Workspaces and name it after the question or project, such as “Office lease renewal” or “Q3 supplier review”.
- 2
Add documents
Add PDFs from your library — your own, your team's or ones shared with you for editing. Free allows 10 documents per workspace, Pro 50, Business 200.
- 3
Label and pin
Give documents short labels such as “Lease 2025”; labels appear in citations. Pin a document to its current version, or let it follow the latest one.
- 4
Ask
Type the question and press Ask, or Ctrl/⌘ + Enter. Untick documents you want left out of this answer.
- 5
Check the sources
Open each numbered citation to see the document, version, page and quote, and jump to the page in the editor.
1Your PDF
2Ask or summarize
3Answers that cite pages
Ask questions that compare
Workspaces are most useful when the answer is spread over several files. “What notice period does each agreement require, and do they differ?” or “Which proposals include onboarding in the price?” get answers that name each document. Each document is searched on its own, so one long file can't crowd out the short ones.
For a single document, Chat with PDF in the editor is quicker. To get a written overview of everything in the workspace instead of one answer, generate a brief.
Why versions matter
If someone edits a document after you asked, a citation still says which version it read. Pin the signed version of an agreement and let the new draft follow the latest version, then ask what the draft changes.
Limits
Scanned PDFs need OCR before they can be searched or cited, and text in images and charts isn't read. Pages beyond your plan's AI page limit are listed as not read. Workspaces belong to the person who created them and can't be shared with a team yet. Document text is sent to OpenAI's API, whose API data-usage terms by default exclude API data from model training.
Tools used in this guide
- Chat with PDFUpload a PDF and ask questions in plain language; every answer points to the pages it is based on.
- PDF to briefUpload one to five PDFs and get a short report with headed sections, where every claim links to the page it came from.
- AI search in PDFDescribe what you're looking for and get the passages that match its meaning, each with a page number.
- OCR PDFUpload a scan, pick its language, and download a PDF that looks the same but lets you search, select and copy the text.
Questions people ask
What if the documents don't answer my question?
The answer says the documents don't cover it and shows no citations. It doesn't fall back to general knowledge.
How many workspaces can I have?
2 on Free, 50 on Pro and unlimited on Business.
Can I turn the documents into something else?
Yes. The same workspace generates a brief (4 AI credits), a slide outline (4 AI credits), a quiz (2 AI credits) or an audio brief (6 AI credits), each with the same citations.
Can an AI agent ask across my workspace?
Yes. With an agent key that has workspace permissions, Claude, Cursor or another MCP client can add documents, ask across them and generate cited briefs.
Are the answers always right?
No. Citations are checked for exact quotes, not for whether the conclusion drawn from them is right. Read the cited passages before you act.