Part 1 of 4 — LlamaIndex 0.12: The Framework That Makes RAG Actually Buildable

If you've read our RAG fundamentals article, you already understand the concept of Retrieval Augmented Generation. You know that RAG = retrieve relevant context + augment the prompt + generate a better answer.

But there's a massive gap between understanding the concept and actually building it. Who handles the chunking? Who manages the vector store? Who orchestrates the retrieval? How do you handle PDFs, CSVs, and SQL databases all at once?

LlamaIndex handles all of that. This is Part 1 of a 4-part series that takes you from zero to a deployed, production-ready RAG app.

The Two Problems LlamaIndex Was Built to Solve

Before we write a single line of code, you need to understand exactly what problem we're solving. There are two fundamental gaps between what LLMs know and what your application actually needs.

LlamaIndex solves both problems by creating a bridge between your live, private data and the LLM. Not by retraining. By retrieval.

The 3-Phase Architecture: Load → Index → Query

LlamaIndex is built around three phases that every RAG application goes through. Understand these three phases and you understand LlamaIndex.

Here's the beauty of this architecture: Phases 1 and 2 run once. You load and index your documents, save the result to disk, and then every query after that goes straight to Phase 3. No re-embedding. No re-indexing. Just instant answers.

Why Use a Framework? What LlamaIndex Does for You

Let's be honest about what it takes to build RAG from scratch without LlamaIndex.

LlamaIndex handles over 160 data loaders, 40+ vector store integrations, automatic chunking strategies, embedding management, prompt templates, memory management, and streaming — all out of the box.

Environment Setup — Everything You Need

Let's get your machine ready. LlamaIndex 0.12 requires Node.js 20.9.0 or higher (the same requirement as Next.js 16, which we'll use in Part 4).

Prerequisites Check

node --version
# Expected: v20.x.x or v22.x.x

# If you need to update, use nvm:
nvm install 22
nvm use 22

Create Your Project

mkdir my-rag-app && cd my-rag-app
npm init -y

Install LlamaIndex 0.12 + Dependencies

# Core framework
npm install [email protected]

# OpenAI integration (LLM + embeddings)
npm install [email protected]

# TypeScript support (recommended)
npm install -D [email protected] @types/[email protected] [email protected]

# dotenv for environment variables
npm install [email protected]

Configure TypeScript

// tsconfig.json
{
  "compilerOptions": {
    "target": "ES2022",
    "module": "ESNext",
    "moduleResolution": "bundler",
    "esModuleInterop": true,
    "strict": true,
    "outDir": "./dist",
    "rootDir": "./src"
  },
  "include": ["src/**/*"]
}

Set Up Your Environment Variables

# .env
OPENAI_API_KEY=sk-proj-...your-key-here...

Project Structure

my-rag-app/
├── src/
│   ├── index.ts          ← Your main entry point
│   └── data/
│       └── sample.txt    ← Your test document
├── .env
├── .gitignore
├── tsconfig.json
└── package.json

Create your first data file:

mkdir -p src/data
echo "LlamaIndex is a data framework for building LLM applications.
It was created by Jerry Liu and Simon Suo in 2022.
The TypeScript version (LlamaIndex.TS) was released in 2023.
LlamaIndex supports over 160 data connectors, 40+ vector stores,
and a comprehensive suite of agentic tools." > src/data/sample.txt

Your First RAG Query — Hello, LlamaIndex

Let's write your first working RAG system. This exact code will load a document, index it, and answer questions about it. Every single line is explained.

// src/index.ts
import "dotenv/config";
import { Settings } from "llamaindex";
import { OpenAI } from "llamaindex/llms";
import { OpenAIEmbedding } from "llamaindex/embeddings";
import { SimpleDirectoryReader } from "llamaindex/ingestion";
import { VectorStoreIndex } from "llamaindex/indices";

async function main() {
  // ── Step 1: Configure your LLM and embedding model ──────────────────
  // Settings is LlamaIndex's global config object. Set it once, use everywhere.
  Settings.llm = new OpenAI({
    model: "gpt-4o", // Most capable OpenAI model
    apiKey: process.env.OPENAI_API_KEY,
    temperature: 0.1, // Low temp = more factual answers
  });

  Settings.embedModel = new OpenAIEmbedding({
    model: "text-embedding-3-large", // Best OpenAI embedding model
    apiKey: process.env.OPENAI_API_KEY,
  });

  // Chunking config: how big each text piece is
  Settings.chunkSize = 512; // tokens per chunk
  Settings.chunkOverlap = 50; // overlap between chunks

  // ── Step 2: Load your documents ──────────────────────────────────────
  // SimpleDirectoryReader reads every file in the folder automatically.
  // Supports .txt, .pdf, .docx, .csv, .json, .md, and more.
  const reader = new SimpleDirectoryReader();
  const documents = await reader.loadData("./src/data");
  console.log(`Loaded ${documents.length} document(s)`);

  // ── Step 3: Build the index ───────────────────────────────────────────
  // This chunks the documents, embeds each chunk, and stores the vectors.
  // Under the hood: document → chunks → OpenAI embeddings → in-memory store.
  console.log("Indexing documents...");
  const index = await VectorStoreIndex.fromDocuments(documents);
  console.log("Index ready!");

  // ── Step 4: Create a query engine and ask a question ─────────────────
  // asQueryEngine() wraps the index with retrieval + synthesis logic.
  const queryEngine = index.asQueryEngine();

  const response = await queryEngine.query({
    query: "Who created LlamaIndex and when was it founded?",
  });

  console.log("\n── Answer ─────────────────────────────");
  console.log(response.toString());
  console.log("────────────────────────────────────────");
}

main().catch(console.error);

Run it:

npx ts-node src/index.ts

Expected output:

Loaded 1 document(s)
Indexing documents...
Index ready!

── Answer ─────────────────────────────
LlamaIndex was created by Jerry Liu and Simon Suo in 2022.
The TypeScript version was released in 2023.
────────────────────────────────────────

What Just Happened? (Under The Hood)

Key Takeaways

Try It Yourself

What's Next in the Series