context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
embedding — A list of numbers representing a piece of text's meaning, so that similar meanings end up numerically close and can be searched.
chunking — Splitting documents into passages small enough to embed and retrieve individually — the step that quietly determines whether retrieval works at all.
reranking — A second pass that re-scores retrieved candidates by reading each one against the query, fixing the ordering that fast vector search got approximately right.
Why it matters
If you're building RAG pipelines, Contextual Retrieval shows how prepending chunk-specific context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition → before embeddingA list of numbers representing a piece of text's meaning, so that similar meanings end up numerically close and can be searched.Full definition → (and BM25 indexing) cuts retrieval failures substantially over naive chunkingSplitting documents into passages small enough to embed and retrieve individually — the step that quietly determines whether retrieval works at all.Full definition → — a practical fix you can implement with existing embedding and rerankingA second pass that re-scores retrieved candidates by reading each one against the query, fixing the ordering that fast vector search got approximately right.Full definition → stacks.
Key quotes
“The problem is that traditional RAG solutions remove context when encoding information, which often results in the system failing to retrieve the relevant information from the knowledge base.”
Anthropic
“This method can reduce the number of failed retrievals by 49% and, when combined with reranking, by 67%.”
Anthropic
“If your knowledge base is smaller than 200,000 tokens (about 500 pages of material), you can just include the entire knowledge base in the prompt that you give the model, with no need for RAG or similar methods.”
Anthropic
“the one-time cost to generate contextualized chunks is $1.02 per million document tokens”
Anthropic
“we found that Reranked Contextual Embedding and Contextual BM25 reduced the top-20-chunk retrieval failure rate by 67% (5.7% → 1.9%).”
