KNOWLEDGE INDEXES
Turn Raw Data Into Retrieval-Ready Knowledge.
Configure how unstructured documents are parsed, chunked, embedded, and mapped into searchable vector-sparse indexes.
01
Document
Raw document stream
02
Parsing
Tables & text extracted
03
Chunking
Semantic boundary split
04
Metadata
Lineage & access tags
05
Embeddings
1536-dim vectors
06
Vector Index
Sub-millisecond query
Primary Workspace Index
Status: ReadyProduct KnowledgeDemo Data
453
Documents
8,420
Chunks
2 min
Last Updated
Chunk Size: 512 tokens (64 token overlap)
Embedding Model: text-embedding-3-small (1536 dims)
Distance Metric: Cosine Similarity
Sparse Algorithm: BM25 with lowercase stemming