Integration · Documents

Connect Claude & ChatGPT to PDF Vector

Ask your AI to read a PDF, answer a question about an invoice or pull the transactions out of a bank statement with PDF Vector. It can extract fields into a JSON schema you define, including from passports and ID cards. It can also search academic papers, read them as text, follow citations and search research grants.

Connect PDF Vector free

Free plan, no credit card. Takes about a minute.

Try asking

“What is the total and due date on this invoice? (link to PDF)”
The AI asks PDF Vector about the invoice and gives you the total and the due date.
“Extract all transactions from this bank statement as a table.”
It extracts the transactions using a prompt and a JSON schema, so you get the same fields for every row.
“Turn this Word file into plain text.”
It parses the document from a public URL or uploaded content and returns readable text.

What your AI can do

Ask and parse documents

Turn PDFs, Word files and images into text, or ask plain-language questions about them.

Structured extraction

Pull fields from invoices, bank statements, ID documents and any other document into a JSON schema you choose.

Academic research

Search papers, read them as markdown, follow citations, find similar papers and search research grants.

“Find recent papers on battery recycling from 2022 onwards.”
It searches academic papers by query and year range and lists titles, authors and IDs.
“Which papers cite this DOI, and what does it reference?”
It lists the citing papers and the references of that paper.
“Suggest sources I could cite for this paragraph.”
It finds academic papers that match the paragraph you paste.

15 tools for PDF Vector

These tools are switched on when you connect. You can switch any of them off, or require your approval before it runs.

ask_bank_statementChanges dataAsk a `question` about a bank statement, such as the closing balance or a specific transaction. Give it as a `url`, `base64` or `document_id`.
ask_documentChanges dataAsk a plain-language `question` about a document and get an answer. Give the document as a `url`, `base64` content or `document_id`.
ask_invoiceChanges dataAsk a `question` about an invoice, such as the total or due date. Give the invoice as a `url`, `base64` or `document_id`.
extract_bank_statement_dataChanges dataExtract transactions or balances from a bank statement using a `prompt` and a JSON `schema`. Give it as a `url`, `base64` or `document_id`.
extract_document_dataChanges dataPull structured data out of a document: pass a `prompt` describing what you want and a JSON `schema` for the result, plus the document as a `url`, `base64` or `document_id`.
extract_id_document_dataChanges dataExtract fields (name, birth date, document number, expiry) from a passport, driving licence or ID card using a `prompt` and a JSON `schema`. Give it as a `url`, `base64` or `document_id`.
extract_invoice_dataChanges dataExtract fields from an invoice (vendor, line items, totals) using a `prompt` and a JSON `schema`. Give the invoice as a `url`, `base64` or `document_id`.
find_citations_for_textChanges dataFind academic papers that could be cited for a `paragraph` of text you provide.
find_similar_papersChanges dataFind papers similar to a given paper `id`.
get_paper_citationsChanges dataList the papers that cite a paper and the papers it references, by paper `id`.
get_papersChanges dataGet full details for academic papers by their `ids` (for example DOIs or IDs returned by search_papers).
parse_documentChanges dataTurn a document (PDF, Word, image and similar) into readable text. Give the file as a public `url`, `base64` content or an existing `document_id`.
read_paperChanges dataTurn an academic paper into readable markdown text, by paper `id` or PDF `url`.
search_grantsChanges dataSearch research grants by `query`, optionally filtered by deadline range and funding amount.
search_papersChanges dataSearch academic papers by `query`, optionally limited by `year_from` and `year_to`. Returns titles, authors and IDs for further lookups.

Set up in three steps

1
Pick the app
Create a free PipMCP account and choose PDF Vector from the app list.
2
Paste your key
Paste a PDF Vector API key. Sign in at app.pdfvector.com, open your instance and go to its Settings tab. Copy the API key there and paste it into PipMCP.
3
Add the link to your AI
You get a personal MCP link. Add it to Claude, ChatGPT or Cursor:
1. Click your name, then Settings › Connectors › Add custom connector.
2. Paste your link as the Remote MCP server URL.
3. Switch it on from the + menu in a chat.

Questions

What can the AI do in PDF Vector?

The AI can parse documents into text, ask questions about documents, invoices and bank statements, and extract structured data from invoices, bank statements, ID documents and other files. It can also search academic papers, read them, list citations, find similar papers, suggest citations for a paragraph and search research grants.

Does the AI see my PDF Vector API key?

No. Your PDF Vector API key is encrypted at rest and never shown to the AI. After you save it, it is not shown again, not even to you. The AI only sees the results of the tools it calls.

Can I control what the AI is allowed to do?

Yes. You choose which tools are switched on, for example only invoice tools or only paper search. Any tool can be set to require your approval before it runs, and every tool call is logged.

Does it work with ChatGPT?

Yes. In ChatGPT go to Settings › Apps & Connectors › Advanced and turn on Developer mode, then add your PipMCP link. Developer mode needs a paid ChatGPT plan: Plus, Pro, Business or Enterprise. The same link also works in Claude (Settings › Connectors › Add custom connector), Cursor and other MCP clients.

What does it cost?

PipMCP has a free plan with no credit card required. Paid plans bill per completed task. You also need your own PDF Vector account.

How do I give the AI a document?

Each document tool takes the file as a public URL, as base64 content or as the ID of a document already in PDF Vector. A link to a PDF the AI can reach is usually the simplest way.