SourceTwin AI

SourceTwin AI finds different file representations of the same real-world document. Select PDFs, scanned PDFs, JPEG, PNG, HEIC, or TIFF images from Files or Photos. SourceTwin creates private working snapshots, then compares embedded text, on-device Vision OCR, page coverage, SHA-256 hashes, and visual fingerprints. It groups likely matches even when filenames or formats differ. Review one candidate group at a time. The suggested minimum safe set preserves every detected page, searchable text source, and visible difference. More than one file may stay when each preserves something useful. Low-confidence and unreviewed files remain included by default, and you make every final decision. Create a separate safe-set folder containing byte-for-byte verified copies and a PDF audit. Selected source files are never renamed, moved, overwritten, or deleted. SourceTwin does not authenticate signatures, certify document completeness, or provide legal advice. Free scans support up to 20 files and 100 pages, including the complete review workflow, safe-set folder, and PDF audit. SourceTwin Pro Lifetime is one non-consumable in-app purchase that supports up to 500 files and 2,000 pages per workspace, known-document memory, ZIP, CSV, JSON, SHA256SUMS, and local result history. There is no subscription. Document analysis runs on device. SourceTwin has no account, ads, analytics, tracking, developer document server, or cloud AI. On supported Apple Intelligence devices, an on-device model may phrase verified evidence; deterministic rules remain authoritative and a local template is always available.