Document Processing
Understanding how Platted processes documents helps you work more effectively with the system.
Processing Pipeline
Each document goes through several stages:
1. Upload
The document file is securely uploaded and stored.
2. Text Extraction
Text is extracted from the document using OCR (Optical Character Recognition) for scanned documents or direct extraction for digital PDFs.
3. Classification
AI analyzes the document to determine:
- Document type (Deed, Mortgage, Lien, etc.)
- Document category (Exam or Supporting)
- Confidence level
4. Data Extraction
AI extracts key information:
- Parties - Grantors, grantees, borrowers, lenders
- Recording Info - Instrument number, book/page, recording date
- Legal Description - Property identification
- Dates - Execution date, effective date
- Amounts - Consideration, loan amounts
5. Matter Identification
The system identifies matters (chain of title entries) from the extracted data.
Processing Status
Documents show their processing status:
| Status | Meaning |
|---|---|
| Uploading | File is being transferred |
| Processing | AI is analyzing the document |
| Classified | Type has been determined |
| Extracted | Data has been pulled |
| Review | Needs manual review |
| Complete | Processing finished |
| Error | Processing failed |
Confidence Scores
The system assigns confidence scores to extractions:
- High (85%+) - Automatic acceptance
- Medium (50-84%) - May need review
- Low (under 50%) - Requires manual review
Low confidence items appear in the Review Queue.
Retry Processing
If a document fails to process:
- Click the document row
- Click Retry Processing
- The document re-enters the pipeline
Split Document Detection
If a document contains multiple distinct records (e.g., a deed and its associated affidavit), the system may flag it for splitting. You'll be prompted to confirm whether to treat it as separate documents.