Agentic RAG
Agentic RAG (Retrieval-Augmented Generation) extends traditional RAG with agent capabilities, enabling multi-step reasoning, dynamic tool use, and context-aware interactions to provide domain-specific answers from curated knowledge bases rather than general knowledge.
- Implementation example: Azure AI guide (Azure Innovation Station) using e-books as knowledge base
- Implementation example: Cole Medin - RAG 2.0 with Knowledge Graphs (RAG 2.0 agentic RAG plus knowledge graphs) for advanced agent search in custom knowledge bases
- Focuses on combining Agentic RAG with knowledge graphs to optimize how agents search and utilize custom knowledge bases
- Requires active Azure subscription for deployment
- Focuses on document-specific retrieval (e.g., e-books) over general knowledge
- Enables agent-like behavior through structured RAG workflows
- Guide for creating an AI agent using Agentic RAG in Azure AI and foundry