# Which frameworks integrate best with Zilliz for agentic RAG?

Source: https://zilliz.com/ai-faq/which-frameworks-integrate-best-with-zilliz-for-agentic-rag
Author: zilliz

LangGraph, LlamaIndex, and OpenAI Agents are the primary 2026 frameworks for building agentic RAG systems; all integrate seamlessly with Zilliz Cloud.

**LangGraph:** Built by LangChain, LangGraph provides a graph-based state machine for multi-step workflows. Agents define states, transitions, and tools. Zilliz integrates as a tool that the agent calls for semantic search.

**LlamaIndex:** Purpose-built for RAG, LlamaIndex provides agent abstractions (ReAct agents, query engines) and built-in vector database integration. Agents can compose multiple query engines and retrieval sources.

**OpenAI Agents:** Use the function-calling API to enable agents to retrieve from Zilliz Cloud. The vector database becomes a callable tool in the agent's tool registry.

**CrewAI:** Orchestrates multi-agent workflows where each agent has access to shared memory stores. Zilliz Cloud provides the shared vector memory that agents collaborate with.

**LangChain:** While not agent-first, LangChain's tool framework works with Zilliz for agent-driven retrieval.

All of these frameworks treat Zilliz as a vector store tool, not a knowledge layer. Agents query Zilliz based on reasoning, not fixed routes. With Zilliz Cloud, teams skip infrastructure management and focus on agent logic.

**Related Resources:**
- [Agentic RAG with Claude and LlamaIndex](https://zilliz.com/blog/agentic-rag-using-claude-3.5-sonnet-llamaindex-and-milvus)
- [Local agentic RAG with LangGraph](https://zilliz.com/blog/local-agentic-rag-with-langraph-and-llama3)
- [Intelligent RAG with LangGraph](https://zilliz.com/blog/build-intelligent-rag-with-langserve-langgraph-and-milvus)
- [What is Zilliz Cloud](https://zilliz.com/what-is-zilliz-cloud)