What is Retrieval-Augmented Generation (RAG)?
Retrieval-augmented generation is a technique where a language model retrieves relevant passages from a specific body of data and answers using them, rather than from its training alone.
Also called: RAG
In more detail
The model is given the source material at query time, so answers can cite where they came from and can cover information the model never saw during training. It also means the data is never absorbed into model weights.
Why it matters
RAG is how enterprises use AI on their own information without training a model on it — cheaper than fine-tuning, updatable the moment a document changes, and auditable because answers point at sources.
How iLeaf approaches it
iLeaf treats retrieval quality as the thing to measure first. Chunking, embeddings and hybrid keyword search are tuned against a graded set of real questions before any answer generation is layered on top.