Release Notes
Improved Document Extraction with Upgraded OCR
Harvey’s new OCR system delivers 40% better accuracy, improved table and handwriting recognition, and reliable support for long documents.
Release Date
Jun 27, 2025
Categories
Knowledge
We’ve rolled out a major upgrade to Harvey’s document processing system, now powered by a significantly improved OCR (Optical Character Recognition) engine. This enhancement boosts the accuracy and reliability of text extraction across all document types.
Expanded capabilities:
- Handwritten text recognition: Accurately reads notes, annotations, and marginalia
- Improved table extraction: Better structure recognition for complex tables and embedded data
- Long document support: Stable performance for documents exceeding 200+ pages
Quality improvements:
- 40% improvement in overall extraction accuracy
- Fewer recognition errors and improved layout parsing
- Enhanced extraction of key fields like names, dates, and numeric values
- Smarter document categorization and tagging