RAG usage on an internal knowledge base

In short, RAG is a simple three-step process: indexing, retrieval, and generation ¹.

^{image source from Medium 1}

The key thing to remember from this diagram is the use of embeddings and a vector database as a proxy to find a relevant context to the query ².

^{image source from Medium 2}

^{a hight level RAG sample architecture 3}

other references

• Retrieval Augmented Generation - RAG
• Choosing between LlamaIndex or LangChain

---

1. How to improve RAG results in your LLM apps: from basics to advanced ↩ ↩²

2. You Can Now Build A Chatbot To Talk To Your Internal Knowledge Base ↩ ↩²

3. Level Up your RAG: Tuning Embeddings on Vertex AI ↩