Xycora's Document Intelligence module ingests documents in any format your clients and opposing parties send — scanned PDFs, native Word and Excel files, email threads, audio transcripts, and even phone photographs of paper documents. Our OCR pipeline achieves over 99% character accuracy on standard court documents and handles handwritten annotations with a confidence score that lets you decide how much human review is warranted before relying on extracted text.
Once ingested, the discrepancy detection engine runs automatically across the entire document set for a matter. It flags internal inconsistencies — a date in a covering letter that does not match the date in the attached contract, a payment figure that differs between a bank statement and an invoice — with a confidence score and an exact page and line reference for both conflicting passages. Early design partners found an average of 14 discrepancies per matter that had not been caught during manual review.
Every finding is pin-cited: Xycora links its analysis output directly to the source passage in the source document, so you can verify any claim in one click. There is no need to trust the AI's summary — the raw evidence is always one click away, and every citation is verified against the stored document hash to ensure it has not been modified since upload.
Supported file formats
Xycora Document Intelligence accepts PDF (native and scanned), Microsoft Word (.docx, .doc), Microsoft Excel (.xlsx, .xls), plain text (.txt), email exports (.eml, .msg), audio transcripts (.vtt, .srt, .txt), and JPEG/PNG photographs. Maximum single-file size is 500 MB. Bulk uploads via drag-and-drop or API support any number of files simultaneously, with processing queued automatically and results delivered to the Case Workspace as each document completes.