Chunking means splitting long documents into small sections so an AI can later find exactly the right passages.
Example
An 80-page manual is split into sections of roughly half a page each.
For pros
Chunk size and overlap strongly affect the quality of a RAG system. Chunks that are too small lose context; chunks that are too large dilute the search. Good methods split at headings or paragraphs and store metadata such as title and page.
