Skip to main content

Overview

The Document Extraction workflow allows you to upload credit agreements and automatically extract structured CDM (Common Domain Model) data using AI-powered analysis.

Step-by-Step Process

1. Access Document Parser

Navigate to the Document Parser tab in the main application interface.

2. Upload Document

Click the upload button and select your credit agreement file. Supported formats:
  • PDF (.pdf)
  • Word documents (.docx)
  • Plain text (.txt)

3. Extraction Process

The system automatically:
  • Extracts text from the document
  • Analyzes content using AI (GPT-4o, vLLM, or HuggingFace)
  • Identifies key entities (parties, facilities, terms)
  • Structures data according to CDM format
  • Validates policy compliance

4. Review Extracted Data

Review the extracted information:
  • Credit Agreement: Main agreement details
  • Parties: Borrowers, lenders, guarantors
  • Facilities: Loan facilities and terms
  • Policy Evaluation: Compliance check results

5. Save to Library

Once validated, save the document to your library for future use in document generation or trade execution.

Extraction Features

AI-Powered Analysis

CreditNexus uses advanced Large Language Models to understand complex legal text:
  • Entity extraction (names, amounts, dates)
  • Relationship mapping (party roles, facility structures)
  • Term identification (interest rates, payment schedules)

CDM Compliance

All extracted data follows FINOS Common Domain Model standards:
  • Standardized data structures
  • Interoperable with other financial systems
  • Policy validation at creation point

Policy Enforcement

Real-time policy evaluation ensures:
  • Regulatory compliance (MiCA, Basel III, FATF)
  • Sanctioned party checking
  • Risk assessment
  • Automated flagging of issues

Tips for Best Results

  1. Document Quality: Ensure documents are clear and readable
  2. Complete Information: Include all relevant pages
  3. Standard Formats: Use standard credit agreement formats when possible
  4. Review Carefully: Always review extracted data before saving

Troubleshooting

Extraction fails or returns incomplete data:
  • Check document format and quality
  • Ensure document is not corrupted
  • Try re-uploading the document
  • Contact support if issues persist
Policy evaluation blocks extraction:
  • Review policy evaluation details
  • Address flagged compliance issues
  • Consult compliance team if needed

Next Steps

After extraction, you can:
  • Use data in Document Generation
  • Create trades from extracted facilities
  • Export data for external systems
  • Share data via FDC3 context