🚀 AskBase - RAG System

Retrieval-Augmented Generation (RAG) system built with Microsoft Kernel Memory and Semantic Kernel. Transform your documents into an intelligent Q&A assistant in minutes.

---

✨ Features

• 🧠 Dual AI Stack - Kernel Memory for RAG + Semantic Kernel for advanced prompting
• 💬 Multi-Session Chat - Conversation history with follow-up question suggestions
• 💾 Flexible Storage - Choose between SimpleVectorDb (zero-config) or Qdrant (high-performance)
• 🎯 Smart Validation - AI-powered answer quality assessment

---

🏗️ Architecture

┌─────────────────────────────────────────────────────────┐
│                   AskBase API (.NET 8)                  │
├─────────────────────────────────────────────────────────┤
│  Controllers  │  ChatController  │  SystemController    │
├─────────────────────────────────────────────────────────┤
│   Services    │  KernelMemory   │  SemanticKernel      │
│               │  SessionManager │  ConversationManager  │
├─────────────────────────────────────────────────────────┤
│  AI Layer     │  Kernel Memory (RAG Pipeline)           │
│               │  Semantic Kernel (Prompt Engineering)   │
├─────────────────────────────────────────────────────────┤
│ Vector Store  │  SimpleVectorDb  │  Qdrant             │
├─────────────────────────────────────────────────────────┤
│   AI Models   │  Azure OpenAI (GPT-4o + Embeddings)    │
└─────────────────────────────────────────────────────────┘

---

📸 Demo

Interactive AI assistant with source attribution and follow-up suggestions

---

🚀 Quick Start

Prerequisites

• .NET 8 SDK
• Azure OpenAI (with GPT-4o and text-embedding-ada-002 deployments)
• Optional: Docker (only if using Qdrant)

Setup

1. Clone the repository

   git clone https://github.com/ramital/AskBase.git
   cd AskBase

2. Configure Azure OpenAI

   Update appsettings.json:

   {
     "OpenAI": {
       "ApiKey": "your-azure-openai-key",
       "Endpoint": "https://your-resource.openai.azure.com/"
     }
   }

3. Add your documents

   # Place documents in the Documents folder
   cp your-files.pdf Documents/

4. Run the application

   dotnet run --project AskBase

5. Open your browser

   http://localhost:5000
   

---

📖 Usage

Web Interface

Navigate to http://localhost:5000 for an intuitive chat interface with:

• Real-time document Q&A
• Conversation history
• AI-suggested follow-up questions
• Source citations

⚙️ Configuration

Vector Storage Options

Choose your vector database in appsettings.json:

SimpleVectorDb (Default - No Docker Required):

{
  "KernelMemory": {
    "VectorStore": "SimpleVectorDb",
    "SimpleVectorDb": {
      "StoragePath": "Documents/.km_vectors"
    }
  }
}

Qdrant (High Performance):

{
  "KernelMemory": {
    "VectorStore": "Qdrant",
    "Qdrant": {
      "Endpoint": "http://localhost:6333"
    }
  }
}

Start Qdrant with Docker:

docker run -d -p 6333:6333 qdrant/qdrant

---

Production Considerations

• Use Azure Managed Identity for OpenAI authentication
• Deploy Qdrant on Azure Container Apps for scale
• Store secrets in Azure Key Vault
• Enable Application Insights for monitoring

---

🛠️ Tech Stack

Component	Technology
Framework	.NET 8 / ASP.NET Core
AI Orchestration	Microsoft Kernel Memory
Prompt Engineering	Microsoft Semantic Kernel
Vector Database	SimpleVectorDb / Qdrant
LLM	Azure OpenAI (GPT-4o)
Embeddings	text-embedding-ada-002

---

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

---

📝 License

MIT License