Data Library versus advanced RAG setup
Key Takeaway
A data library is a preconfigured quick start for RAG. It needs Data 360 enabled and holds one data source. The advanced setup needs Data 360 implemented and fits combining sources, real-time or zero-copy data, and harmonized customer data.
Retrieval Augmented Generation (RAG) grounds an AI response in your own data. Agents can get RAG in two ways. The Agentforce Data Library is a preconfigured quick start. The advanced setup, which you build directly in Data 360, takes more work but gives you more control and more data options.
How RAG works
RAG works in three steps:
- It retrieves relevant information from a knowledge store.
- It augments the original prompt with that information.
- The LLM generates a response.
RAG lets you bring proprietary, up-to-date data to the LLM without retraining or fine-tuning the model.