Your file stays on your device
The text is extracted from the PDF in your browser, and only that text is sent to answer your question. The file itself is never uploaded — you can confirm there's no file upload in DevTools → Network.
Privacy & Security
Mostly — and the honest details matter. Your file never leaves your device: the text is extracted in your browser and only that text is sent, never the file. DockDocs doesn't store your document or the conversation on its servers. The one caveat: to answer, that text has to pass through an AI model, so a truly “zero data sent” AI isn't possible.
“AI document chat” sounds like uploading your file, but it doesn't have to be. Here's the real data flow.
The text is extracted from the PDF in your browser, and only that text is sent to answer your question. The file itself is never uploaded — you can confirm there's no file upload in DevTools → Network.
The extracted text and the conversation are processed for your request and not retained on DockDocs servers. (If you explicitly save a chat with My Chats, that's your own saved copy — opt-in, not automatic.)
To generate an answer, the extracted text passes through an AI model. DockDocs can't promise what a model provider does with data in transit, so a truly “zero data sent” AI isn't possible. What you can rely on: the file never leaves your device, DockDocs keeps nothing, and only the minimum — text — is sent.
Open DevTools → Network (F12) and ask a question. You'll see a small request carrying text (your question and the relevant extracted text) — not a large upload carrying your file. If you ever saw the file itself being uploaded, that would be a different, server-upload design.
DockDocs is built to send the minimum and keep nothing — and to be honest about the part it can't control.
Only the extracted text needed to answer is sent; your original PDF never leaves your device.
The request is processed and not stored on DockDocs servers. Saving a conversation is opt-in (My Chats), never automatic.
When the AI answers, DockDocs shows the source quote from your document when it can locate it — and tells you when it can't. It won't invent a citation.
No. The text is extracted from the PDF in your browser and only that text is sent to answer your question — the file itself is never uploaded. You can confirm there's no file upload in DevTools → Network.
The AI request is processed and not retained on DockDocs servers. The one exception is opt-in: if you explicitly save a chat with the My Chats feature, that's your own saved copy — it never happens automatically.
No — generating an answer requires the extracted text to pass through an AI model, so a truly zero-data AI isn't possible today. The honest goal is to send the minimum (text, not your file) and not retain it. DockDocs is built that way.
The extracted text goes to the configured AI model provider to generate the answer, and DockDocs doesn't store it. As with any AI tool, avoid pasting secrets you wouldn't want a model provider to process.
Use the client-side tools — compress, merge, split, redact and similar — which run entirely in your browser and send nothing. AI features inherently send the extracted text, because a model has to read it to answer.