Task

Currently the Agentic Sample workflow has Legal and Support Agents https://github.com/thepetk/langgraph-agents/blob/main/examples/langchain-langgraph/workflow.py from the Quickstart RAG notebook dir https://github.com/rh-ai-quickstart/RAG/tree/dev/notebooks 

We need to ensure vector store optimization in the Agentic Sample workflow i.e. specific vector store looked up after classification rather than all of them. For example, if a query is determined/classified as a Legal domain, then only search the Legal Vector Store rather than all of the Vector Stores.

(The idea is also to have Llama Stack Client implementation for predominantly all the Agents, except one Agent can be OpenAI client for raising awareness)

Background

Dependencies and Blockers

QE impacted work
Documentation impacted work 

Acceptance Criteria