AI research workspace for your documents

Ask your documents.
Get answers you
can verify.

Upload research papers and reports, then have a conversation with them. Querex retrieves the relevant passages and answers with page-level citations — grounded, streamed, and traceable.

Free to use · sign in with Google, GitHub, or email

Q3-Financial-Report.pdf
Grounded
What drove the revenue increase this quarter?
Searching Q3-Financial-Report.pdf

Growth was led by enterprise renewals, with average deal size rising 12% quarter-over-quarter across the segment.

Q3-Financial-Report.pdf · page 14

Grounded in your sources

Answers come from your files, not the model's memory.

Page-level citations

Every claim links to the exact page it came from.

Real-time streaming

Answers appear token by token — no waiting.

Multi-document synthesis

Ask one question across many files at once.

Cited by page, not by vibe

Every answer is grounded in your documents and points to the exact page it came from — trace any claim back to its source in one click.

Streams in real time

Token-by-token responses over server-sent events — no spinners, no waiting for the full reply.

Compares across documents

Select multiple PDFs and ask one question across all of them — synthesis, not just retrieval.

Remembers the thread

Pinned documents and persistent sessions — pick up any research conversation exactly where you left it, days later.

Not another AI chatbot

Most tools either chat without proof, or read PDFs without memory. Querex is built to do both — grounded answers, kept in context.

Generic AI chatbot

General-purpose conversation

Page-level source citations
Real-time token streaming
Multi-document synthesis
Admin dashboard & audit log
Automatic retention controls

Typical single-doc PDF tool

Chat with one file at a time

Page-level source citations
Real-time token streaming
Multi-document synthesis
Admin dashboard & audit log
Automatic retention controls
This is Querex

Querex

Grounded research workspace

Page-level source citations
Real-time token streaming
Multi-document synthesis
Admin dashboard & audit log
Automatic retention controls

How it works

01

Upload

Drop in a PDF — it's parsed page by page and embedded into a private vector index.

02

Ask

Ask in plain language. Retrieval finds the passages that actually answer your question.

03

Verify

Every answer streams in with a page citation, so you can check it against the source.

Built in the open

No walled garden, no black box. The full source — retrieval pipeline, auth, admin panel, and all — is on GitHub for anyone to read, audit, or run themselves.

Fully open source

The complete codebase is public. Read every line, or self-host your own instance.

Covered by automated tests

Auth, retention, admin access, and RAG contracts run in CI on every change.

Security-hardened by default

httpOnly sessions, rate limiting, and brute-force protection — not bolted on later.

Stop skimming. Start asking.

Upload your first PDF and get a cited answer in under a minute.