Write apps/ai_agent/docs/concepts-rag-search-architecture.md explaining how /index/message and /search together form a retrieval pipeline: what gets embedded, how OpenAI is used (embedding model, and separately the chat model), and how this pipeline is meant to be consumed by callers.
Acceptance criteria:
- The indexing pipeline (message text → embedding → Weaviate storage) is documented step by step
- The retrieval pipeline (query → embedding → Weaviate similarity search → results) is documented step by step
- Explicitly documents whether/how retrieved results currently feed back into any other endpoint's LLM prompt — if no integration exists yet, the doc states that plainly
Write
apps/ai_agent/docs/concepts-rag-search-architecture.mdexplaining how/index/messageand/searchtogether form a retrieval pipeline: what gets embedded, how OpenAI is used (embedding model, and separately the chat model), and how this pipeline is meant to be consumed by callers.Acceptance criteria: