Multi-Document Contradiction Checker
Upload 2 or more related PDFs or paste their text, and an AI model will compare them against each other for names, dates, amounts and addresses that don't match between documents - flagged as things worth double-checking, not confirmed errors.
Drag & drop files, or click to browse
Add 2 or more related PDFs (e.g. different versions or copies of the same paperwork)
Questions fréquentes
How is this different from the Document Inconsistency Checker?
The Document Inconsistency Checker looks for mismatches within a single document. This tool compares two or more separate but related documents against each other - for example a lease and its renewal, or a resume and a cover letter - and flags facts that don't agree between them.
Does this guarantee it finds every contradiction?
No. This is an AI-assisted pattern-flagging tool, not exhaustive error detection. It can miss real contradictions and can occasionally flag something that's actually fine. Treat every result as "worth a second look," never as a confirmed list of every mismatch between your documents.
What kinds of contradictions does it look for?
The same person's name spelled or formatted differently across documents, dates for what looks like the same event or deadline that conflict, amounts or numbers referring to the same thing that don't match, and addresses that differ when they seem to refer to the same place.
What does "found in" mean in a result?
It names which of your uploaded documents (by filename, for PDFs, or by the label you gave it, for pasted text) the mismatch was spotted between, so you can go check those specific documents.
Can I mix uploaded PDFs and pasted text?
Not in the same check - switch between the "Upload PDFs" and "Paste text" tabs depending on what you have. If you need to include both a PDF and some other text, you can extract the PDF's text elsewhere and paste both in the "Paste text" tab instead.
Is my document uploaded anywhere?
For a PDF, text is extracted in your browser first. Either way, only the plain text (not the original files) is sent to a free AI model to run the comparison, and it isn't stored afterward.