Retrieval-Augmented Generation enhances an LLM by retrieving relevant information from external knowledge sources and incorporating that material into the context supplied to the model during generation. This permits a pretrained model to answer questions using organization-specific information without requiring its underlying model parameters to be retrained for every knowledge update.
Cisco describes RAG as a technique that connects LLM applications to additional data sources, including internal repositories containing sensitive customer or business information. Cisco's enterprise RAG guidance specifically emphasizes using proprietary organizational data to ground responses in enterprise information. Therefore, providing access to proprietary data is the correct answer.
Option A is incorrect because RAG does not eliminate ingestion, parsing, chunking, cleaning, embedding, or indexing requirements. Option B is too absolute: grounding can improve factual relevance but does not inherently remove model bias. Option D describes modification of model parameters, which is characteristic of training or fine-tuning rather than standard RAG retrieval.
Cisco explicitly includes RAG as a required DCAI workload type, so candidates must distinguish retrieval-based augmentation from model training and fine-tuning.
Study Guide Reference: 1.0 AI Fundamentals and Applications — 1.1.a RAG.
===============
