Anugana.Rag 🚀

Cross-Platform Retrieval-Augmented Generation (RAG) AI Assistant

Powered by Uno Platform, .NET 10, Qdrant Vector Database, and OpenRouter AI.

---

🌟 Overview

Anugana.Rag is a modern, high-performance, cross-platform AI desktop and mobile application that brings document intelligence directly to your device. Upload PDFs, text files, or markdown documents into your personal vector knowledge base and converse with an AI assistant that answers questions using exact context snippets from your documents.

---

🧠 What is an Embedding Model & Why is it Used?

An Embedding Model is an AI model that converts human text into a mathematical array of numbers called a Vector (e.g. [0.024, -0.158, 0.892, ...]).

How Embeddings Power RAG:

1. Semantic Meaning: Embeddings capture the conceptual meaning of words, not just exact keyword matches. For example, the text "automobile" and the query "car" produce vectors that are mathematically very close.
2. Indexing (Knowledge Base): When you upload a document, Anugana.Rag breaks it into text chunks and sends each chunk to the Embedding Model to generate a vector. These vectors are saved into Qdrant Vector Database.
3. Retrieval (Chat): When you ask a question, your query is converted into a vector by the same Embedding Model. Qdrant compares your query vector against all document vectors to instantly find the most relevant text snippets for the LLM to read and answer from.

---

🔑 How to Get FREE Access Keys & Free Models

Anugana.Rag works out of the box with 100% Free Tier Cloud Providers so you can start querying your documents without spending a dime!

1. 🗄️ Free Qdrant Vector DB Cloud (Free Forever 1GB Cluster)

Qdrant offers a generous Free Forever cloud cluster with 1GB storage (enough for tens of thousands of document vectors):

1. Sign up at https://cloud.qdrant.io.
2. Click Create Cluster → choose the Free Tier (1GB RAM).
3. Once created, copy your Cluster URL (e.g. https://xxx.cloud.qdrant.io) and API Key.
4. Paste them into Anugana.Rag Settings under Qdrant Endpoint URL and API Key.

2. 🤖 Free OpenRouter API Key & Free Models

OpenRouter provides access to top-tier LLMs and embeddings with completely free models:

1. Sign up at https://openrouter.ai.
2. Go to Keys → click Create Key (no credit card required).
3. Copy your API Key and paste it into Anugana.Rag Settings under OpenRouter API Key.

🆓 Free AI Chat Models (Set in Settings under Chat Model ID):

• openrouter/free (Auto-selects the best available free model)
• nvidia/nemotron-3-ultra-550b-a55b:free
• google/gemma-4-31b-it:free
• google/gemma-4-26b-a4b-it:free
• openai/gpt-oss-20b:free
• poolside/laguna-m.1:free

🆓 Free Embedding Models (Set in Settings under Embedding Model ID):

• nvidia/llama-nemotron-embed-vl-1b-v2:free
• nvidia/nemotron-3-embed-1b:free
• openai/text-embedding-3-small (High precision default option)

---

📱 App Feature Guide & Screenshots

Anugana.Rag features an intuitive 3-tab navigation bar at the bottom:

💬 1. RAG Inference Chat

• Ask Anything: Type your query into the input bar and press Enter to send (or Shift + Enter for a new line).
• Thinking Indicator: Watch the 🤔 Searching knowledge base & thinking... indicator as your vector DB is queried.
• Real-Time Markdown: Answers stream in live with code blocks, bold text, lists, and formatted headings.
• Copy Response: Click the 📋 Copy button on any response card for instant clipboard copy (✅ Copied!).
• Source Citations: Every answer lists the exact source file, page number, and vector relevance score (📌 manual.pdf | Page 2 | Relevance: 92%).

---

📄 2. Knowledge Base (Document Ingestion)

• Multi-File Picker: Click 📁 Browse & Select Files (Multi-Select) to choose multiple .pdf, .txt, or .md files at once.
• Automatic Page Chunking: Documents are split into overlapping text chunks, embedded, and stored into Qdrant automatically.
• Real-Time Progress: Track processing status for each document in your batch (⏳ [1/3] document.pdf: Indexing vectors...).
• Database Stats: View total indexed vector count and connection status at a glance.

---

⚙️ 3. Settings & Configuration

• App Theme Switcher: Toggle between ☀️ Light Mode, 🌙 Dark Mode, and 💻 System Default.
• Vector DB & AI Endpoints: Configure Qdrant URLs, API keys, and test connections with instant feedback badges (✅ Connection Successful!).
• RAG Search Parameters: Fine-tune Chunk Size (500), Overlap (50), Top-K Results (4), Score Threshold (0.50), and System Prompts.

---

🛠️ Getting Started & Local Development

Prerequisites

• .NET 10 SDK
• Qdrant Cluster (Cloud or Local Docker)
• OpenRouter API Key

Building and Running Locally

1. Clone the Repository:

   git clone https://github.com/avikeid2007/Anugana.Rag.git
   cd Anugana.Rag

2. Run Desktop App:

   dotnet run --project Anugana.Rag/Anugana.Rag.csproj -f net10.0-desktop

3. Run WebAssembly Web App:

   dotnet run --project Anugana.Rag/Anugana.Rag.csproj -f net10.0-browserwasm

---

📦 Downloads & Live Links

Access Anugana.Rag via live web demo, official app stores, or direct GitHub release binaries:

Platform	Link / Download	Description
🌐 Web App	anugana.avnishkumar.co.in	Run directly in browser via WebAssembly (No install needed).
🪟 Microsoft Store	Get on Microsoft Store	Official Windows Store app package.
📱 Google Play Store	Get on Google Play	Official Android app from Google Play.
🪟 Windows Standalone	Anugana.Rag-Windows-Standalone.exe	Portable single-file executable from GitHub Releases.
🪟 Windows MSIX	Anugana.Rag-Windows-Setup.msix	Native Windows installer package from GitHub Releases.
📱 Android APK	Anugana.Rag-Android.apk	Direct Android APK package from GitHub Releases.

---

📄 License & Privacy

• License: This project is licensed under the MIT License - see the LICENSE.txt file for details.
• Privacy Policy: Read our Privacy Policy for details on local data storage and API communications.