RAG (Retrieval-Augmented Generation) solves the problem of the model not knowing your private information: documents are chunked, vectorized and stored in a knowledge base; at question time, the most relevant chunks are retrieved and passed to the model along with the question, so it answers from real material instead of guessing from training data.
Enclave's character knowledge bases are RAG in action: upload material to a character, and it looks things up before answering — fewer hallucinations, and it can answer with specifics from your domain.
Try it yourself in Enclave
Open it in your browser — no credit card, no install.