Glossary
RAG (Retrieval Augmented Generation)
An AI technique that improves LLM accuracy by retrieving relevant information from your own knowledge base before generating a response — letting the model “look things up” instead of relying only on training data.
General definition
RAG, or Retrieval Augmented Generation, is an AI technique that enhances large language model accuracy by retrieving relevant information from your own knowledge base before generating a response. Rather than relying solely on what it learned during training, the model can “look things up” at answer time.
The process runs in four steps:
- Query processing — the user submits a question
- Information retrieval — the system searches external sources such as documents, databases or websites
- Context integration — retrieved information is combined with the user’s query
- Response generation — the LLM answers using both its training and the retrieved data
Documents are converted into vector embeddings — numerical representations that capture semantic meaning — so the system can automatically pull only the most relevant information for each query.
RAG in the Ethora ecosystem
Ethora positions RAG as essential for building business-grade AI chat bots and knowledge assistants. Because responses are grounded in source material, RAG dramatically reduces hallucinations and keeps answers accurate, current and transparent.
The AI SDK provides several ways to implement it — an Unstructured Docs Indexer for document ingestion, a Business Website Crawler for pulling in site content, and integration with the Embeddable AI Chat Widget and AI Agents SDK so your bots answer from your own knowledge.
