Document Prompt Chain — Multi-Step Analysis Pipeline

I will build a document processing prompt chain — covering ingestion and pre-processing, section identification and classification, per-section targeted extraction, cross-section synthesis, quality validation, and structured output — enabling accurate analysis of complex documents that exceed what a single prompt can handle reliably within its context window. Long and complex documents processed with a single prompt suffer from attention dilution — the model's focus is spread across the entire document, reducing precision on specific sections; a chained approach that directs focused attention at each section individually and then synthesises findings produces significantly higher extraction accuracy.

The chain covers document segmentation, per-segment classification, targeted extraction prompts for each segment type, synthesis step combining segment-level extractions, validation against your schema, and structured output delivery.