Explaining RAG in under 5 minutes

What is retrieval-augmented generation? What are embeddings? How can you use this to supplement an existing LLM with your own data? Let's talk about it.