Feature Overview
Implement Agentic RAG with Llama Stack, with capabilities for:

• multilanguage RAG
• vision retrieval
• interaction with granite 3.x vision model

Goals

• Basic Agentic RAG flow
• Multi-turn interaction before and after retrieval
• English, non-English/multilingual
• Using vectordb over Llama Stack
  • Vectordb should support multimodal RAG

• Interaction with granite 3.x vision model

Stretched goal:

• Hybrid Search/Hybrid Retrieval

Requirements

• Implementation of Agentic RAG using Llama Stack API
  • Llama Stack Client SDK
  • Llama Stack Tool RAG Provider

Done

Use Cases
<your text here>

Out of Scope
<your text here>

Documentation Considerations
<your text here>

Questions to Answer
<your text here>