Home / Glossary / Chunking
What is Chunking?
Chunking splits long documents into smaller pieces that fit comfortably in an LLM context window and are independently retrievable. osFoundry auto-chunks notes, KBs, and uploaded files as they’re created.
Detail
A naive split-every-N-tokens chunking loses context. Better strategies preserve semantic boundaries — split on paragraphs, headings, or sentence boundaries; optionally overlap chunks; optionally enrich each chunk with surrounding context (e.g. parent heading).
Chunk size affects retrieval quality. Smaller chunks are more precise but lose context; larger chunks have more context but worse retrieval signal. Common range: 200-1000 tokens with 10-20% overlap.
How osFoundry approaches Chunking
osFoundry’s chunking pipeline auto-runs on document creation. Strategy is configurable in the chunking station — pick semantic, structural, or fixed-size; tune size and overlap.
Related terms
Related features