Upload your PDFs, ask a question, then click the answer to see the exact sentence, table row, or figure it came from, highlighted on the page.
See it in action:
You have to cite your source
Down to the sentence, table row or figure.
You have a lot of PDFs
More than any chat window will take.
You re-upload the same PDFs
Every new session, from scratch.
You fill in long questionnaires
Upload the template. Ask your AI to fill it in. Done.
You compare two versions
Two contracts, two specs, two reports.
You use more than one AI
One PDF library, whichever assistant you open.
Research papers, contracts, reports, manuals. Drop them in.
Only you can access your data.
Every PDF is converted, chunked, and embedded into a searchable knowledge base, automatically.
Claude, ChatGPT, Perplexity, Gemini... no matter what you use. It's simpler than it sounds. We'll walk you through it step by step.
You're ready to start chatting with your PDFs!
Upload an empty template. Ask your AI to fill it in. Done.
Every line it writes carries a link that opens the source document at the exact spot the answer came from: a sentence, a table row, or a figure.
There's no fixed limit. Index a handful of PDFs or thousands, they're all available in every session.
Think of MCP as a USB port for AI: a standard way for tools like Claude, ChatGPT, Gemini, and Cursor to plug into outside data, instead of you copy-pasting everything in by hand. pdf2mcp is one of those connections, giving you direct access to your PDFs without the need to reupload them.
No. Upload once, and every future session can query your PDF library automatically, no re-uploading required.
Three things, and the first is the big one. Every answer links back to the exact sentence, table row, or figure it came from, highlighted on the page, so you can check it instead of trusting it. There's no limit on how many PDFs you index. And your library works in every conversation you start, on any MCP-compatible assistant, rather than being locked inside one chat.
Any MCP-compatible AI assistant, including Claude, ChatGPT, Perplexity, Gemini, Cursor, and Copilot Studio. If it speaks MCP, it works, and we walk you through the setup for each one.
Password-protected or encrypted PDFs aren't supported. Everything else works.
Yes. Every answer carries a link that opens the document at the exact sentence, table row, or figure the answer was drawn from, highlighted on the page, so you can check it yourself instead of just trusting the AI.
No account, no setup. Connect your AI client in a couple of minutes.
Uploads are anonymous and deleted after 48 hours.