Skip to content

Build your first RAG application

This example indexes a local document directory and answers an Arabic question. It uses Chroma locally, OpenAI for generation, and a local embedding model.

Prepare credentials

$env:OPENAI_API_KEY = "your-key"

export OPENAI_API_KEY="your-key"

Create a knowledge base

Put PDFs, DOCX files, text files, or supported images under ./knowledge-base, then create app.py:

import os

from Muffakir import MuffakirRAG

rag = MuffakirRAG(
    data_dir="./knowledge-base",
    llm_provider="openai",
    llm_model="gpt-4.1-mini",
    api_key=os.environ["OPENAI_API_KEY"],
    embedding_model="sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2",
    vector_db_provider="chroma",
    db_path="./muffakir_db",
    language="ar",
)

result = rag.ask("ما هي الخطوات المطلوبة لتقديم طلب الإجازة؟")
print(result["answer"])

Run it with python app.py.

What happens in the pipeline?

  1. Muffakir parses and cleans the source documents.
  2. It splits the content into chunks and stores embeddings in the configured vector store.
  3. It retrieves relevant chunks for the question.
  4. The configured LLM receives the retrieved context and returns an answer.

Use rag.get_similar_documents(query, k=5) when you only need the retrieved documents. Use rag.get_similar_documents_with_trace(...) when you want retrieval details for debugging.

Next steps