One master document,
compared against every document in your list.
Upload a master — a template, a policy, last year's contract — then a list of documents to check against it. Each one is compared separately, and each gets its own report you can read or download. Four documents means four reports, not one blended summary.
Access is approved by the site owner. Requesting it just submits your email address — you'll be able to sign in once it's approved. No password to create: you receive a short code by email each time you sign in.
What each comparison tells you
Semantic diff
Every substantive change as a table: the item, the value it had, the value it has now, and what the change amounts to. Never a new figure without its original.
Contradictions
Where the two documents make incompatible claims — with each document's exact quoted wording, a severity, and why it matters in practice.
Side-by-side
The comparable fields lined up per document, with a similarity score so you can see at a glance how far apart they are.
Redline
Word-level tracked changes — additions underlined, deletions struck through. Computed exactly, with no model involved, so it is perfectly reproducible.
How it works
- Request access with your email address, and wait for approval.
- Sign in with the code emailed to you — there is no password.
- Upload one master document and the list to compare it against.
- Collect each comparison's report separately, or the whole run as one archive.
Your documents stay on this server
Every comparison runs on models hosted on this machine, and scanned documents are read here too. Nothing — not the document, not a page of it, not the text pulled out of it — is sent to an outside AI service. That is the point when the documents are contracts, policies or anything you would not paste into a website.