PolicyVector 1.4.1 improves duplicate matching during import, updates PDF processing dependencies, and fixes an error when checking Local AI status.
More reliable duplicate matching
- Different documents with generic names such as “Chapter 1” or “Transmittal Memorandums” are no longer treated as the same Miscellaneous policy simply because their names match. PolicyVector uses authoritative policy identifiers when available and file contents for documents without them.
- Import review no longer borrows revision details from an unrelated Miscellaneous document. Before replacing a saved file, PolicyVector checks the incoming and saved PDFs again so an outdated preview cannot authorize the wrong replacement. Identical files still qualify to be skipped.
Safer PDF processing
- Updates the PDF reader to address security issues in malformed PDFs that could cause excessive memory use or leave processing stuck. The PDF rendering engine is also updated for continued compatibility.
- Updates the file-locking dependency used by the local model cache. These maintenance updates preserve your existing library, outlines, and settings.
Clearer Local AI failures
- If a Local AI status check fails, PolicyVector reports the runtime as unavailable instead of retrying the same failing call without its timeout. This prevents the retry from hiding the original failure or causing another error.
This Windows release covers changes since Windows 1.4.0. macOS and Mobile updates are distributed separately.