Chunk retrieval
Retrieving individual segments of a document rather than whole pages, which is how most AI search and RAG systems select what to feed a…
GEO AIContext length
Context window
The maximum amount of text a language model can consider at once, which bounds how much retrieved content can be supplied when generating an answer.
In full
Even with large windows, retrieval systems are selective because relevance degrades and cost rises with volume, so only a handful of passages usually reach the model. This is why concise, self-contained, information-dense sections outperform sprawling pages in retrieval contexts: they fit, and they survive selection. Window size also constrains how much of a long document an agent will actually read before answering.
Example
A 12,000-word guide is represented in an answer by two retrieved sections; the rest never reaches the model.
Related terms
Retrieval-augmented generation
An architecture where a system retrieves relevant documents or passages first and then has a language model generate an answer conditioned…
Semantic chunking
Splitting documents along meaning boundaries — sections, topics, complete ideas — rather than at fixed character counts, to produce…
Knowledge cutoff
The date after which a language model has no training data, so anything more recent is unknown to it unless supplied through retrieval.
Grounding
Constraining a language model's output to information retrieved from a specified source — a search index, a document set or a database —…