A document index stores processed representations of documents, including extracted text, metadata, keyword indexes, and vector embeddings. Retrieval queries the index rather than the original files.
Index construction occurs during ingestion and must be updated as documents are added.
Alternative Names:
Search Index, Retrieval Index
Why it Matters?
Index scope defines what the system can find, and gaps are silent. Documents that failed OCR, files in unsupported formats, and material added after the last index update are invisible to search regardless of relevance. In litigation this creates a real risk, since a party relying on AI search to confirm that no responsive document exists may be relying on an index that never contained it. Verifying ingestion completeness is part of any defensible process.
Frequently Confused with
Related terms
Frequently asked questions
What can make documents invisible to search?
Why does index completeness matter for defensibility?
