This guide provides detailed step-by-step workflows for getting the most out of the Content Processing Solution Accelerator. These "golden path" workflows represent the most common and effective use cases for the solution.
The golden path workflows are designed to:
- Demonstrate the full capabilities of the solution
- Provide a structured learning experience
- Showcase best practices for document processing
- Help users understand the confidence scoring and validation features
- Solution deployed and validated successfully
- Sample schemas registered (Invoice schema)
- Authentication configured
-
Access the Web Interface
- Navigate to your deployed web app URL
- Log in using your configured authentication
-
Select Invoice Schema
- In the Processing Queue pane, select "Invoice" from the schema dropdown
- Verify the schema shows as available
-
Upload Sample Invoice
- Click "Import Content" button
- Select an invoice file from the sample data (PDF, PNG, or JPEG)
- Click "Upload" to submit
-
Monitor Processing
- Watch the file status change from "Uploaded" → "Processing" → "Completed"
- This typically takes 1-2 minutes
-
Review Extracted Data
- Click on the completed file to open the review interface
- Examine the extracted data in the "Extracted Results" tab
- Compare with the source document in the "Source Document" pane
-
Validate and Modify Results
- Edit any incorrect data in the JSON output
- Add notes in the "Comments" section
- Pay attention to confidence scores for each field
-
Save and Approve
- Click "Save" to store your modifications
- Review the process steps in the "Process Steps" tab
- ✅ Invoice data accurately extracted (vendor, amounts, dates, line items)
- ✅ Confidence scores above 80% for most fields
- ✅ Any low-confidence fields flagged for manual review
- ✅ Process steps show successful extraction, mapping, and evaluation
- Invoice workflow completed successfully
- Property Loss Damage Claim Form schema registered
-
Switch to Property Claims Schema
- Select "Property Loss Damage Claim Form" from the schema dropdown
-
Upload Property Damage Document
- Import a property claim form from the sample data
- Monitor the processing workflow
-
Validate Complex Extraction
- Review extracted claim details, damages, and policy information
- Note how the system handles form fields vs. free text
-
Test Validation Features
- Modify extracted data to test validation rules
- Add detailed comments about damage assessments
-
Process Multiple Documents
- Upload additional property claim documents
- Compare extraction accuracy across different document formats
- ✅ Complex form data accurately extracted
- ✅ Multi-modal content (text, images, tables) processed correctly
- ✅ Validation rules applied appropriately
- Basic workflows completed
- Understanding of your specific document types
-
Create Custom Schema
- Follow the Custom Schema Guide
- Define your document structure and required fields
-
Register Your Schema
- Use the schema registration scripts
- Validate schema is available in the web interface
-
Test with Sample Documents
- Start with 2-3 representative documents
- Process and review initial results
-
Refine Extraction Quality
- Analyze confidence scores and accuracy
- Modify schema definitions if needed
- Re-test with updated schema
-
Scale to Production
- Process larger document batches
- Establish quality thresholds
- Set up automated workflows using the API
- ✅ Custom schema accurately processes your document types
- ✅ Confidence scoring helps identify manual review needs
- ✅ Workflow scales to handle production volumes
- Process different document types in the same session
- Compare extraction approaches across schemas
- Understand when to use different processing strategies
- Use programmatic APIs for document submission
- Implement webhook callbacks for processing notifications
- Build custom validation and approval workflows
- Upload multiple documents simultaneously
- Monitor batch processing status
- Export results for downstream systems
- Always review low-confidence extractions manually
- Use comments to document validation decisions
- Track accuracy improvements over time
- 90-100%: High confidence, likely accurate
- 70-89%: Medium confidence, review recommended
- Below 70%: Low confidence, manual review required
- Use consistent document formats when possible
- Ensure good image quality for scanned documents
- Batch similar document types for better consistency
- Check document quality and formatting
- Verify schema matches document structure
- Review and update system prompts if needed
- Reduce document file sizes
- Check Azure quota availability
- Monitor system logs for errors
- Verify app registration configuration
- Check user permissions and role assignments
- Review authentication provider settings
After completing these golden path workflows:
-
Explore Advanced Features
- Custom validation rules
- Webhook integrations
- Batch processing APIs
-
Integrate with Your Systems
- Connect to downstream databases
- Set up automated workflows
- Implement custom business logic
-
Scale Your Solution
- Monitor performance metrics
- Optimize for your specific use cases
- Plan for production deployment
- Technical Documentation: API Guide
- Troubleshooting: Common Issues
- Sample Data: Download samples
- Community: Submit issues
This guide is based on the automated test suite golden path workflows that validate the core functionality of the solution.