Build RAG Chatbot with Haystack, Zilliz Cloud, Mistral Codestral Mamba, and HuggingFace all-MiniLM-L12-v1

Introduction to RAG

Retrieval-Augmented Generation (RAG) is a game-changer for GenAI applications, especially in conversational AI. It combines the power of pre-trained large language models (LLMs) like OpenAI’s GPT with external knowledge sources stored in vector databases such as Milvus and Zilliz Cloud, allowing for more accurate, contextually relevant, and up-to-date response generation. A RAG pipeline usually consists of four basic components: a vector database, an embedding model, an LLM, and a framework.

Key Components We'll Use for This RAG Chatbot

This tutorial shows you how to build a simple RAG chatbot in Python using the following components:

Haystack: An open-source Python framework designed for building production-ready NLP applications, particularly question answering and semantic search systems. Haystack excels at retrieving information from large document collections through its modular architecture that combines retrieval and reader components. Ideal for developers creating search applications, chatbots, and knowledge management systems that require efficient document processing and accurate information extraction from unstructured text.
Zilliz Cloud: a fully managed vector database-as-a-service platform built on top of the open-source Milvus, designed to handle high-performance vector data processing at scale. It enables organizations to efficiently store, search, and analyze large volumes of unstructured data, such as text, images, or audio, by leveraging advanced vector search technology. It offers a free tier supporting up to 1 million vectors.
Mistral Codestral Mamba: A code-focused AI model designed to streamline software development by generating, analyzing, and optimizing code. Leveraging the Mamba architecture, it excels at processing long sequences efficiently, offering fast inference and precise code suggestions. Ideal for developers seeking productivity gains in code completion, debugging, and automation tasks.
HuggingFace all-MiniLM-L12-v1: A compact sentence embedding model designed to convert text into dense vector representations for semantic understanding. It balances speed and efficiency with strong performance in tasks like semantic search, text clustering, and retrieval-augmented generation. Ideal for applications requiring low-latency inference or resource-constrained environments while maintaining robust semantic analysis capabilities.

By the end of this tutorial, you’ll have a functional chatbot capable of answering questions based on a custom knowledge base.

Note: Since we may use proprietary models in our tutorials, make sure you have the required API key beforehand.

Step 1: Install and Set Up Haystack

import os
import requests

from haystack import Pipeline
from haystack.components.converters import MarkdownToDocument

from haystack.components.preprocessors import DocumentSplitter
from haystack.components.writers import DocumentWriter

Step 2: Install and Set Up Mistral Codestral Mamba

To use Mistral models, you need first to get a Mistral API key. You can write this key in:

The api_key init parameter using Secret API
The MISTRAL_API_KEY environment variable (recommended)

Now, after you get the API key, let's install the Install the mistral-haystack package.

pip install mistral-haystack

from haystack_integrations.components.generators.mistral import MistralChatGenerator
from haystack.components.generators.utils import print_streaming_chunk
from haystack.dataclasses import ChatMessage
from haystack.utils import Secret

generator = MistralChatGenerator(api_key=Secret.from_env_var("MISTRAL_API_KEY"), streaming_callback=print_streaming_chunk, model='open-codestral-mamba')

Step 3: Install and Set Up HuggingFace all-MiniLM-L12-v1

Haystack'sHuggingFaceAPITextEmbedder can be used to embed strings with different Hugging Face APIs:

The component uses a HF_API_TOKEN environment variable by default. Otherwise, you can pass a Hugging Face API token at initialization with token – see code examples below. The token is needed:

If you use the Serverless Inference API, or
If you use Inference Endpoints.

Here, in this tutorial, we'll use the Free Serverless Inference API. Let's install and set up the model.

To use this API, you need a free Hugging Face token. The Embedder expects the model in api_params.

from haystack.components.embedders import HuggingFaceAPITextEmbedder
from haystack.utils import Secret
from haystack.components.embedders import HuggingFaceAPIDocumentEmbedder
from haystack.dataclasses import Document

text_embedder = HuggingFaceAPITextEmbedder(api_type="serverless_inference_api",
                                           api_params={"model": "sentence-transformers/all-MiniLM-L12-v1"},
                                           token=Secret.from_token("<your-api-key>"))
document_embedder = HuggingFaceAPIDocumentEmbedder(api_type="serverless_inference_api",
                                              api_params={"model": "sentence-transformers/all-MiniLM-L12-v1"},
                                              token=Secret.from_token("<your-api-key>"))

Step 4: Install and Set Up Zilliz Cloud

pip install --upgrade pymilvus milvus-haystack

from milvus_haystack import MilvusDocumentStore
from milvus_haystack.milvus_embedding_retriever import MilvusEmbeddingRetriever

document_store = MilvusDocumentStore(connection_args={"uri": ZILLIZ_CLOUD_URI, "token": ZILLIZ_CLOUD_TOKEN}, drop_old=True,)
retriever = MilvusEmbeddingRetriever(document_store=document_store, top_k=3)

Step 5: Build a RAG Chatbot

Now that you’ve set up all components, let’s start to build a simple chatbot. We’ll use the Milvus introduction doc as a private knowledge base. You can replace it your own dataset to customize your RAG chatbot.

url = 'https://raw.githubusercontent.com/milvus-io/milvus-docs/refs/heads/v2.5.x/site/en/about/overview.md'
example_file = 'example_file.md'
response = requests.get(url)
with open(example_file, 'wb') as f:
    f.write(response.content)
file_paths = [example_file]  # You can replace it with your own file paths.

indexing_pipeline = Pipeline()
indexing_pipeline.add_component("converter", MarkdownToDocument())
indexing_pipeline.add_component("splitter", DocumentSplitter(split_by="sentence", split_length=2))
indexing_pipeline.add_component("embedder", document_embedder)
indexing_pipeline.add_component("writer", DocumentWriter(document_store))
indexing_pipeline.connect("converter", "splitter")
indexing_pipeline.connect("splitter", "embedder")
indexing_pipeline.connect("embedder", "writer")
indexing_pipeline.run({"converter": {"sources": file_paths}})

# print("Number of documents:", document_store.count_documents())
  
question = "What is Milvus?"  # You can replace it with your own question.

retrieval_pipeline = Pipeline()
retrieval_pipeline.add_component("embedder", text_embedder)
retrieval_pipeline.add_component("retriever", retriever)
retrieval_pipeline.connect("embedder", "retriever")

retrieval_results = retrieval_pipeline.run({"embedder": {"text": question}})

# for doc in retrieval_results["retriever"]["documents"]:
#     print(doc.content)
#     print("-" * 10)

from haystack.utils import Secret
from haystack.components.builders import PromptBuilder

retriever = MilvusEmbeddingRetriever(document_store=document_store, top_k=3) 
 text_embedder = HuggingFaceAPITextEmbedder(api_type="serverless_inference_api",
                                           api_params={"model": "sentence-transformers/all-MiniLM-L12-v1"},
                                           token=Secret.from_token("<your-api-key>"))

prompt_template = """Answer the following query based on the provided context. If the context does
                     not include an answer, reply with 'I don't know'.\n
                     Query: {{query}}
                     Documents:
                     {% for doc in documents %}
                        {{ doc.content }}
                     {% endfor %}
                     Answer: 
                  """

rag_pipeline = Pipeline()
rag_pipeline.add_component("text_embedder", text_embedder)
rag_pipeline.add_component("retriever", retriever)
rag_pipeline.add_component("prompt_builder", PromptBuilder(template=prompt_template))
rag_pipeline.add_component("generator", generator)
rag_pipeline.connect("text_embedder.embedding", "retriever.query_embedding")
rag_pipeline.connect("retriever.documents", "prompt_builder.documents")
rag_pipeline.connect("prompt_builder", "generator")

results = rag_pipeline.run({"text_embedder": {"text": question}, "prompt_builder": {"query": question},})
print('RAG answer:\n', results["generator"]["replies"][0])

Optimization Tips

As you build your RAG system, optimization is key to ensuring peak performance and efficiency. While setting up the components is an essential first step, fine-tuning each one will help you create a solution that works even better and scales seamlessly. In this section, we’ll share some practical tips for optimizing all these components, giving you the edge to build smarter, faster, and more responsive RAG applications.

Haystack optimization tips

To optimize Haystack in a RAG setup, ensure you use an efficient retriever like FAISS or Milvus for scalable and fast similarity searches. Fine-tune your document store settings, such as indexing strategies and storage backends, to balance speed and accuracy. Use batch processing for embedding generation to reduce latency and optimize API calls. Leverage Haystack's pipeline caching to avoid redundant computations, especially for frequently queried documents. Tune your reader model by selecting a lightweight yet accurate transformer-based model like DistilBERT to speed up response times. Implement query rewriting or filtering techniques to enhance retrieval quality, ensuring the most relevant documents are retrieved for generation. Finally, monitor system performance with Haystack’s built-in evaluation tools to iteratively refine your setup based on real-world query performance.

Zilliz Cloud optimization tips

Optimizing Zilliz Cloud for a RAG system involves efficient index selection, query tuning, and resource management. Use Hierarchical Navigable Small World (HNSW) indexing for high-speed, approximate nearest neighbor search while balancing recall and efficiency. Fine-tune ef_construction and M parameters based on your dataset size and query workload to optimize search accuracy and latency. Enable dynamic scaling to handle fluctuating workloads efficiently, ensuring smooth performance under varying query loads. Implement data partitioning to improve retrieval speed by grouping related data, reducing unnecessary comparisons. Regularly update and optimize embeddings to keep results relevant, particularly when dealing with evolving datasets. Use hybrid search techniques, such as combining vector and keyword search, to improve response quality. Monitor system metrics in Zilliz Cloud’s dashboard and adjust configurations accordingly to maintain low-latency, high-throughput performance.

Mistral Codestral Mamba optimization tips

Optimize Mistral Codestral Mamba in RAG by fine-tuning on domain-specific data to align outputs with context, adjusting temperature (lower for precision, higher for creativity), and trimming retrieved documents to relevant chunks to reduce noise. Use quantization for faster inference, enable dynamic batching, and limit max tokens to prevent overlong responses. Cache frequent queries, precompute embeddings for common contexts, and leverage hardware acceleration (e.g., CUDA cores). Monitor latency and accuracy to balance speed and quality, and apply prompt engineering with clear instructions to guide context integration.

HuggingFace all-MiniLM-L12-v1 optimization tips

To optimize the all-MiniLM-L12-v1 model in a RAG setup: preprocess input data by cleaning and normalizing text (lowercasing, removing special characters) to improve embedding quality. Use batch inference for embedding generation to maximize GPU utilization. Fine-tune the model on domain-specific data via contrastive learning to enhance retrieval relevance. Reduce vector dimensionality via PCA if storage or latency is critical. Cache frequently accessed embeddings to minimize recomputation. Quantize the model with Hugging Face’s transformers library for faster inference with minimal accuracy loss. Regularly benchmark performance against your retrieval metrics (e.g., recall@k) to validate optimizations.

By implementing these tips across your components, you'll be able to enhance the performance and functionality of your RAG system, ensuring it’s optimized for both speed and accuracy. Keep testing, iterating, and refining your setup to stay ahead in the ever-evolving world of AI development.

RAG Cost Calculator: A Free Tool to Calculate Your Cost in Seconds

Estimating the cost of a Retrieval-Augmented Generation (RAG) pipeline involves analyzing expenses across vector storage, compute resources, and API usage. Key cost drivers include vector database queries, embedding generation, and LLM inference.

RAG Cost Calculator is a free tool that quickly estimates the cost of building a RAG pipeline, including chunking, embedding, vector storage/search, and LLM generation. It also helps you identify cost-saving opportunities and achieve up to 10x cost reduction on vector databases with the serverless option.

Calculate your RAG cost now.

Calculate your RAG cost

What Have You Learned?

By diving into this tutorial, you’ve unlocked the power of building a modern RAG system using cutting-edge tools! You’ve seen how Haystack acts as the backbone, simplifying the orchestration of your pipeline by seamlessly connecting components like Zilliz Cloud—a blazing-fast vector database that stores and retrieves embeddings at scale. With Mistral Codestral Mamba, you harnessed a lean yet powerful LLM to generate accurate, context-aware responses, while HuggingFace’s all-MiniLM-L12-v1 embedding model transformed raw text into meaningful vector representations, bridging the gap between unstructured data and machine-understandable formats. Together, these tools form a robust RAG pipeline that’s efficient, scalable, and ready to tackle real-world challenges like question answering, content generation, or semantic search. You even learned optimization tricks—like tuning chunk sizes and leveraging Zilliz Cloud’s indexing strategies—to squeeze every drop of performance out of your system. And let’s not forget the free RAG cost calculator introduced here, a game-changer for budgeting resources and scaling wisely without surprises.

Now it’s your turn to take these insights and run with them! Whether you’re enhancing customer support bots, building intelligent search engines, or experimenting with AI-driven content tools, you’ve got the blueprint. Experiment with different models, tweak parameters, and explore how small optimizations can lead to big wins. The world of RAG is evolving rapidly, and you’re already at the forefront. So fire up your IDE, spin up that Zilliz Cloud instance, and start creating. The future of intelligent applications is in your hands—build something awesome, share your wins, and keep pushing the boundaries of what’s possible. Let’s go! 🚀

Further Resources

🌟 In addition to this RAG tutorial, unleash your full potential with these incredible resources to level up your RAG skills.

How to Build a Multimodal RAG | Documentation
How to Enhance the Performance of Your RAG Pipeline
Graph RAG with Milvus | Documentation
How to Evaluate RAG Applications - Zilliz Learn
Generative AI Resource Hub | Zilliz

We'd Love to Hear What You Think!

We’d love to hear your thoughts! 🌟 Leave your questions or comments below or join our vibrant Milvus Discord community to share your experiences, ask questions, or connect with thousands of AI enthusiasts. Your journey matters to us!

If you like this tutorial, show your support by giving our Milvus GitHub repo a star ⭐—it means the world to us and inspires us to keep creating! 💖