Source Document

Source document is the original file or document uploaded to DocLD for processing. Parsing extracts text, tables, and layout from the source; chunking and embedding operate on that extracted content. Extraction and RAG work from the parsed result, not the raw file.

Lifecycle

1. Document upload — Source document is uploaded
2. Parse — Parsing extracts content; OCR may run for scanned or image content
3. Chunk and embed — Content is chunked and embedded for vector search
4. Extract / RAG — Extraction and RAG use the parsed content

Related Concepts

Source documents enter via document upload. Parsing and OCR process them; chunking and embedding prepare content for vector search and RAG.