Ingestion

Ingestion is the process of bringing documents into DocLD and making them available for search or processing. It typically includes upload, parsing, chunking, embedding, and storing chunks in the vector index (e.g., for a knowledge base). Once ingested, documents can be queried via vector search and RAG.

Pipeline

Upload → Parse → Chunk → Embed → Upsert to vector index

Batch processing can ingest many documents; each may have a job. Adding documents to a knowledge base triggers ingestion for those documents. Reindex re-runs ingestion (e.g., after changing chunking or embedding settings).

Related Concepts

Ingestion is the full flow from document upload to document processing and indexing. Reindex repeats ingestion. Knowledge base search depends on completed ingestion.