Uploading documents
What Nyrava accepts, how OCR works, and how to structure large productions.
Supported file types
- PDF (searchable or scanned)
- Microsoft Word (.docx)
- Plain text (.txt) and Markdown (.md)
- Images with legible text (PNG, JPG, TIFF)
- CSV — parsed as tabular data
OCR behavior
Scanned PDFs and images pass through OCR automatically. Extraction quality is very good on typed documents and moderate on legible handwriting. Very poor scans should be re-scanned or reviewed before extraction is trusted.
Batching large productions
Productions over a few hundred documents should be uploaded in batches. Each batch triggers incremental indexing so downstream engines can start earlier. There is no hard cap on total corpus size, but per-batch uploads should stay under roughly 500 documents for responsive UX.
Labeling & bates
Keep filenames descriptive and include bates ranges when applicable. Nyrava uses filenames as part of the citation surface, so SMITH000123-000145_Dep_Chen.pdf resolves cleaner than doc7.pdf.