RAG (Retrieval-Augmented Generation)
A technique that combines LLMs and document databases for responses based on corporate sources.
RAG (Retrieval-Augmented Generation) is a technique that enriches an LLM's responses with information retrieved from a corporate vector database. It allows for building chatbots that respond only based on their own documents, reducing hallucinations and updating knowledge without the need for re-training.
Practical examples
- Support chatbots for corporate manuals
- Legal assistants for judicial rulings
- Q&A on technical documentation
- Semantic search on knowledge bases