LlamaIndex

Architecture Overview

LlamaIndex provides a layered architecture: data connectors ingest documents from various sources, indexes organize them for efficient retrieval (vector, keyword, knowledge graph), and query engines expose them to agents. The AgentRunner manages the agent loop, maintaining a TaskState and StepHistory. Agents use a ReAct-style loop with tool abstractions that wrap query engines, APIs, and other data sources. The Workflows API adds event-driven orchestration for complex pipelines.

When to Use LlamaIndex

• Knowledge management and document Q&A systems
• Agentic RAG with multi-step retrieval
• Data-driven agents over private enterprise data
• Multi-document analysis and comparison
• Structured data extraction from unstructured sources

Strengths & Weaknesses

Quick Start

from llama_index.core.agent import ReActAgent
from llama_index.llms.openai import OpenAI
from llama_index.core.tools import FunctionTool

def multiply(a: float, b: float) -> float:
    """Multiply two numbers and return the result."""
    return a * b

def add(a: float, b: float) -> float:
    """Add two numbers and return the result."""
    return a + b

# Wrap functions as tools
multiply_tool = FunctionTool.from_defaults(fn=multiply)
add_tool = FunctionTool.from_defaults(fn=add)

# Create a ReAct agent
llm = OpenAI(model="gpt-4o")
agent = ReActAgent.from_tools(
    [multiply_tool, add_tool],
    llm=llm,
    verbose=True,
)

response = agent.chat("What is (3 + 5) * 2?")
print(response)

Features at a Glance

Notable Users

Resources