Cache augmented generation is an alternative to Retrieval augumented generation (RAG)

Comparison to RAG

in RAG: The model retrieves or is retrieved ideally the exact chunk of relevant data to work with.
In CAG: The model stores all of the information in its context window.

Pros and Cons:

• CAG does not require an retrieval pipeline
• CAG depends on the LLM with

Resource on the comparison: https://www.montecarlodata.com/blog-rag-vs-cag/