A Generative AI Engineer at a home appliance company has been asked to design an LLM based application that accomplishes the following business objective: answer customer questions on home…

Question

Options

• ASplit instruction manuals into chunks and embed into a vector store. Use the question to retrieve
• BCreate an interaction matrix of historical user questions and appliance instruction manuals. Use
• CCalculate averaged embeddings for each instruction manual, compare embeddings to user query
• DUse an LLM to summarize all of the instruction manuals. Provide summaries of each manual and

• A
  82% (37)
• B
  4% (2)
• C
  11% (5)
• D
  2% (1)

Explanation

This is a textbook Retrieval-Augmented Generation (RAG) use case. The correct high-level pipeline is: (1) chunk instruction manuals into manageable passages, (2) embed each chunk and index the embeddings in a vector store, (3) at query time embed the user's question, retrieve the most relevant chunks via similarity search, (4) pass retrieved chunks plus the question to an LLM to generate a grounded answer. Option B (interaction matrix) is a collaborative-filtering approach suited to recommendation systems, not Q&A over documents. Option C (averaged embeddings per manual) discards chunk-level detail needed to answer specific questions. Option D (summarizing all manuals upfront) compresses away the precise details customers ask about and doesn't scale to many documents.

No community discussion yet for this question.