Augmentation is the critical bridge between Retrieval and Generation. In this stage, the system takes the user's original query and "augments" it with the relevant context retrieved from the vector database.
The process of constructing a comprehensive prompt for the Large Language Model (LLM) by combining the User Query, Retrieved Context, and System Instructions. This ensures the LLM has all the necessary facts to generate an accurate, grounded answer.
The original question or task provided by the user.
The most relevant text chunks found in the Vector Database during the retrieval phase.
Instructions telling the LLM how to behave (e.g., "Only use the provided context to answer").