While LLMs (Large Language Models) address generic knowledge very well, it is the combination of LLMs with internal documents that delivers results that can be trusted.
This so-called RAG (Retrieval Augmented Generation) approach consists of a preparation phase whereby documents are uploaded, converted, segmented, vectorized and indexed.
Once a question is asked a retriever leveraging semantic vectors will select snippets (segments) that are closest to the question while the LLM generates an answer based on these snippets.
Please fill in the form below to access our Whitepaper (PDF format).