Chunking

Chunking is how AI systems split content into smaller passages, typically 200 to 500 tokens, before indexing and retrieving it, meaning a page is rarely evaluated as one continuous whole but as a set of independent segments competing separately for retrieval.

Why it matters

Chunking is the underlying reason self-contained sections matter more than a page's overall narrative flow: if a passage doesn't make sense pulled out of its surrounding context, it's less likely to be retrieved or cited well, regardless of how good the full page reads top to bottom.

Related terms

• Passage Retrieval

  Passage retrieval is the step where an AI system selects specific relevant text chunks, not entire pages, to build its answer, meaning a page can be retrieved for one passage while the rest of it never enters the AI system's consideration for that query at all.

• Answer Block / Atomic Answer

  An answer block, or atomic answer, is a self-contained passage, typically 40 to 100 words, that fully answers a specific question on its own, without requiring a reader or an AI system to reference surrounding paragraphs to understand it.

• Extractability

  Extractability is how cleanly a chunk of content can be lifted and reused by an AI system without needing surrounding context to remain accurate or coherent, functioning as a practical measure of whether a passage would still make sense if it were the only thing quoted.

• Retrieval-Augmented Generation (RAG)

  Retrieval-Augmented Generation (RAG) is the architecture behind most AI search: the system retrieves a set of relevant documents or passages at query time, then generates its answer grounded in that retrieved material, rather than purely from what it learned during training.