Automated Document Processing Pipeline

# Automated Document Processing Pipeline

You are a document automation engineer. Design a complete AI pipeline that processes [DOCUMENT_TYPE] documents arriving in [INPUT_SOURCE] without human intervention.

## Ingestion Stage

Define how the agent:
- Monitors [WATCH_FOLDER_OR_EMAIL] for new documents
- Accepts formats: PDF, DOCX, images (via OCR), or [ADDITIONAL_FORMATS]
- Validates that each document meets minimum quality standards before processing

## Extraction Stage

For each document, extract the following structured fields:
[FIELD_LIST]

Specify the extraction confidence threshold — flag documents where confidence falls below [CONFIDENCE_THRESHOLD]% for human review.

## Classification Stage

Classify each document into one of these categories:
[DOCUMENT_CATEGORIES]

Route classified documents to the appropriate destination:
- Category A → [DESTINATION_A]
- Category B → [DESTINATION_B]
- Unclassified → [REVIEW_QUEUE]

## Post-Processing

After successful extraction and classification:
- Write the structured data to [DATABASE_OR_SPREADSHEET]
- Archive the original document in [ARCHIVE_LOCATION] with a timestamp and document ID
- Trigger a downstream action in [CONNECTED_SYSTEM] if the document matches [TRIGGER_CONDITION]

## Error Handling

Log all failures to [ERROR_LOG]. Alert [ALERT_RECIPIENT] if more than [ERROR_THRESHOLD] documents fail within [TIME_WINDOW].