Assisted Chat RAG

Retrieval-Augmented Generation (RAG) database generator for OpenShift Container Platform Assisted Installer documentation.

What is RAG?

Retrieval-Augmented Generation (RAG) is a powerful AI technique that combines the capabilities of large language models with external knowledge retrieval. Instead of relying solely on the model's training data, RAG systems:

1. Retrieve relevant information from a knowledge base (vector database)
2. Augment the user's query with this retrieved context
3. Generate more accurate and up-to-date responses using both the retrieved information and the model's capabilities

This approach enables AI systems to provide more accurate, contextual, and current information by leveraging external documents and knowledge sources.

Documentation Source

This RAG system processes documentation from Red Hat's Assisted Installer for OpenShift Container Platform. The source material is the official PDF documentation:

• Source: Red Hat Documentation - Assisted Installer for OpenShift Container Platform 2025
• Format: Converted from PDF to Markdown using docling

The processed documentation is stored in the docs/ directory as a markdown file, making it easier to chunk and embed for the RAG system.

Prerequisites

• Python 3.12
• uv (Python package manager)
• Git with Git LFS support
• Podman (for container builds)

Installation

1. Clone the repository:

   git clone https://github.com/carbonin/assisted-chat-rag
   cd assisted-chat-rag

2. Install dependencies:

   make install

   This will:

   • Install Python dependencies using uv
   • Set up Git LFS for large file support

Generating the RAG Database

Quick Start

To generate the complete RAG database with default settings:

make build-db

This single command will:

1. Install all dependencies
2. Download the embedding model
3. Process the documentation
4. Create the vector database

Step-by-Step Process

If you prefer to run the process step by step:

1. Download and convert documentation (optional - docs are already included):

   make docs

   This downloads the latest PDF from the official docs site and converts it to markdown.

2. Download the embedding model:

   make all-mpnet-base-v2

   This downloads the sentence-transformers/all-mpnet-base-v2 model for creating embeddings.

3. Build the RAG database:

   uv run ./main.py \
     -o llama_stack_vector_db \
     -f docs \
     -md all-mpnet-base-v2 \
     -mn sentence-transformers/all-mpnet-base-v2 \
     -i ocp-assisted-installer-2025-07-18 \
     --vector-store-type llamastack-faiss

Configuration Options

The RAG database generation supports various configuration options:

• -o: Output directory for the vector database
• -f: Input folder containing markdown documents
• -md: Directory containing the embedding model
• -mn: Name of the embedding model
• -i: Index name for the vector store
• --chunk: Chunk size for document processing (default from lightspeed-rag-content)
• --vector-store-type: Type of vector store (llamastack-faiss)

Output

The generated RAG database will be saved in the llama_stack_vector_db/ directory.

Container Image

Build a container image for deployment:

make build-image

The default image name is quay.io/carbonin/assisted-chat-rag:latest. You can override this:

IMAGE_NAME=your-registry/your-image:tag make build-image

Cleaning Up

Remove generated files:

make clean

This removes:

• llama_stack_vector_db/ directory
• all-mpnet-base-v2/ directory

Help

View all available commands:

make help