answer_question
├── retrieve
│   ├── embed_query
│   └── vector_search
├── build_prompt
└── llm_call
