This Langflow flow creates a document processing system that extracts structured data from contracts and generates SQL insert statements. The flow takes a file as input and processes it through an AI agent that has been specifically instructed to extract contracting company information. The agent uses OpenAI's GPT-4o model and follows a detailed prompt template that guides it to identify key business details like legal names, addresses, and representative information.

The system begins with file input components that can handle multiple document formats. For higher-fidelity text extraction that preserves tables and layout, you can use specialized parsers like Docling or Unstructured. The document content flows into a prompt template that provides clear extraction instructions and defines variables for the language model to process.

The extracted information gets processed through a structured output component that enforces a predefined schema with eight specific fields: legal_name, legal_document, business_address, email, phone, representative_name, representative_id, and representative_address. This component uses the same language model to ensure the data conforms to the expected structure and data types. The structured output component acts as a validation layer that converts the agent's response into a consistent JSON format.

The final output displays the processed results through a chat interface that shows the structured data extraction results. The flow combines file reading capabilities with AI-powered information extraction and data validation to create a complete document processing pipeline. Additional data operations and parsing components can clean and format fields before saving to CSV or JSON files, pushing data via API requests, or displaying results in chat outputs.

For deployment and automation, the flow can be triggered through API endpoints, webhook integrations, or programmatic file uploads. This system would be useful for automating the extraction of company information from legal contracts and converting that information into database-ready formats.

The flow can be extended significantly using other Langflow components. For enhanced document parsing, Docling integration preserves reading order, headings, and table structures while exporting to Markdown or HTML formats. Batch processing capabilities allow the model to run across multiple documents simultaneously. For improved accuracy on large document collections, you can implement retrieval-enhanced extraction by storing document chunks in vector databases and retrieving relevant context during processing. Webhook triggers enable event-driven processing from forms and queues, while API integrations send structured results to downstream business intelligence tools and databases.