This project provides an Enterprise RAG Solution Accelerator for securely enabling Azure OpenAI at scale, targeting organizations moving from POC to production. It offers grounded, secure, and scalable ChatGPT-style and Q&A experiences using Azure Cognitive Search and Azure OpenAI, streamlining LLM integration without costly fine-tuning.

How It Works

GPT-RAG implements the Retrieval-Augmented Generation pattern with a modular architecture. It comprises Data Ingestion for optimized chunking and indexing, an Orchestrator (selectable between Semantic Kernel functions or AutoGen agents) for retrieval and response generation, and a Backend for Frontend application interface (React or Chainlit). This approach ensures data grounding, security via Zero-trust principles, and scalability for enterprise use.

Quick Start & Requirements

• Install/Run: Uses Azure Developer CLI (azd).
• Prerequisites: Azure Developer CLI, PowerShell 7+ (Windows), Git, Node.js 16+, Python 3.11, Azure AI services resource.
• Setup: Multiple deployment options available: Basic Architecture (quick demos, no network isolation) and Zero-Trust Architecture (production, network isolation). Zero-Trust involves VM setup and further azd commands.
• Docs: User & Admin Guide, Deployment Instructions

Highlighted Details

• Supports both functional (Semantic Kernel) and agentic (AutoGen) orchestrators.
• Offers React and Chainlit front-end options.
• Provides Basic and Zero-Trust deployment architectures.
• Emphasizes Zero-trust security and Responsible AI principles.

Maintenance & Community

• Active community support via Discord Channel.
• Clear contribution guidelines available in CONTRIBUTING.md.

Licensing & Compatibility

• Project is licensed under the MIT License.
• Microsoft trademarks are subject to brand guidelines; third-party trademarks follow their respective policies.

Limitations & Caveats

The Zero-Trust deployment requires manual steps on a provisioned VM, including installing PowerShell and updating azd. Uploading documents requires the Storage Blob Data Contributor role.