When the web tool is not enough
A web tool is right for a person comparing two documents. It is wrong for a system that receives fifty revised documents a day. At that point the comparison should happen where the documents already are: in the document management system, the PLM, the contract lifecycle tool or the loan origination platform.
One call, three outputs
The API on this site takes a multipart POST with the older PDF, the newer PDF and an output option. It returns a JSON body with a summary (insertions, deletions, formatting changes), a change list where each entry has a kind, a level, the old and new text and page numbers, plus the redlined PDF and the side-by-side HTML as base64. Ask for JSON only when you need the change list alone; it is the fastest option because nothing is rendered.
Integration patterns
- Release gate: on every new version, compare against the previous one and block the release if the change list contains anything outside the sections named in the change request.
- Returned-document check: when a signed or returned document arrives, compare it against the issued version and route anything with substantive changes to a human.
- Audit trail: store the JSON change list and the redlined PDF against the record so the review can be reproduced later.
- Surveillance: poll an external source for new versions and compare each against the last one seen, which is how the engine was first used for FDA labels.
Limits and deployment
The hosted API accepts documents up to 300 pages and 20 MB per file per request, with a 120-second limit. Larger documents, higher volume and documents that cannot leave your network are handled by a private deployment inside your VPC or on-premises, where those limits do not apply. Keys are issued to enterprise accounts and can be rotated at any time.
Try it on your own documents. Upload two versions on the homepage. Documents up to 30 pages need no account, and nothing is stored.