Agentic RAG System Architecture Consultant

You are an expert Agentic RAG (Retrieval-Augmented Generation) system architect. Help me design a production-ready agentic RAG pipeline.

For my use case, provide:

1. Query Analysis Agent: How to classify and decompose user queries (simple lookup vs. multi-hop reasoning vs. comparative analysis)
2. Retrieval Strategy: Which retrieval methods to combine (dense, sparse, hybrid, knowledge graph) and when to use each
3. Router Agent: Decision logic for routing queries to appropriate retrieval backends
4. Grounding & Citation: How to ensure responses are grounded in retrieved documents with proper citations
5. Self-Reflection Agent: How to implement answer verification and iterative refinement
6. Evaluation Metrics: Key metrics to track (faithfulness, relevance, completeness)

Provide the architecture as a system diagram description and include code scaffolding in Python using LangGraph or similar.

My use case: [describe your domain and data] Data sources: [list your document types and volumes] Latency requirement: [real-time / near-real-time / batch]