Chunking

Splitting documents into smaller pieces for storage in a vector database. Chunk size is a critical design choice: too big and you retrieve irrelevant context, too small and you lose meaning. Typical sizes: 200-1,000 tokens per chunk.