Update README with RAG system details and setup

Expanded the AI Tools section to highlight RAG (Retrieval-Augmented Generation) features, including context-aware answers and document embeddings. Added setup instructions for enabling RAG, quick start steps, and new documentation links. Clarified production deployment steps to include Supabase and knowledge base indexing. Enhanced feature lists to reflect RAG capabilities and updated environment variable configuration.

Author: DongDuong2001

README.md

@@ -270,20 +270,32 @@ The build step runs Next.js static analysis, type-checking, and route bundling.
 
 ## AI Tools
 
-The **AI Tools** feature (`/dashboard/ai-tools`) provides an intelligent development assistant running 100% locally with Ollama.
+The **AI Tools** feature (`/dashboard/ai-tools`) provides an intelligent development assistant with **RAG (Retrieval-Augmented Generation)** - AI that knows your platform inside-out!
 
 ### Features
 
+- **🧠 RAG-Enhanced AI** – AI powered by your documentation, features, and codebase
 - **🤖 Smart AI Assistant** – Code generation, debugging, architecture decisions, and technical guidance
-- **🔒 Complete Privacy** – All processing happens on your machine, zero data sent to cloud
-- **💰 Zero Cost** – No API fees, unlimited usage
+- **📚 Context-Aware** – Answers based on your actual platform documentation
+- **🔒 Complete Privacy** – All processing and data stays on your infrastructure
+- **💰 Zero Cost** – No API fees, unlimited usage with local Ollama
 - **🌐 Offline Capable** – Works without internet connection
 - **💬 Modern Chat UI** – User/AI avatars, message bubbles, copy-to-clipboard, typing indicators
-- **📊 Real-time Status** – Shows Ollama connection status
+- **📊 Real-time Status** – Shows Ollama connection status and RAG usage
 - **🧹 Clear Chat** – Reset conversation anytime
 - **📝 Message Counter** – Track conversation length and character count
 
-### Setup
+### What is RAG?
+
+**RAG (Retrieval-Augmented Generation)** combines semantic search with AI generation:
+
+1. **Your Question** → Converted to embedding vector
+2. **Vector Search** → Finds relevant docs from your knowledge base
+3. **AI Generation** → Ollama uses the context to answer accurately
+
+**Result:** AI that actually knows your platform, features, and documentation!
+
+### Quick Start
 
 #### 1. Install Ollama
 
@@ -305,25 +317,52 @@ ollama pull codellama         # Code-specialized (3.8GB)
 ollama pull qwen2.5-coder:7b  # Excellent for programming
 ```
 
-#### 3. Start Using
+#### 3. Enable RAG System
+
+See **[docs/RAG_QUICKSTART.md](docs/RAG_QUICKSTART.md)** for setup:
+
+```powershell
+# 1. Run SQL schema in Supabase
+# 2. Index your knowledge base
+pnpm run index-knowledge
+
+# 3. Test it!
+# Visit http://localhost:3000/dashboard/ai-tools
+```
+
+#### 4. Start Using
 
 ```powershell
 pnpm dev
 # Navigate to http://localhost:3000/dashboard/ai-tools
 ```
 
-See [docs/OLLAMA_SETUP.md](docs/OLLAMA_SETUP.md) for detailed setup instructions.
+See [docs/OLLAMA_SETUP.md](docs/OLLAMA_SETUP.md) for detailed Ollama setup.  
+See [docs/RAG_SYSTEM.md](docs/RAG_SYSTEM.md) for complete RAG documentation.
+
+### RAG Features
+
+- **📄 Document Embeddings** – All docs indexed with semantic vectors
+- **🔍 Smart Search** – Finds relevant content via cosine similarity
+- **⚡ Fast Retrieval** – < 10ms search with pgvector indexes
+- **🎯 Category Filtering** – Search by documentation, features, code, etc.
+- **🔄 Easy Updates** – Re-index when docs change
+- **📊 Similarity Scores** – See how relevant each context is
 
 ### Production Deployment
 
 For production, run Ollama on a separate server:
 
 1. **Deploy Next.js** on Vercel (or any platform)
 2. **Setup Ollama server** on VPS (AWS EC2, DigitalOcean, Hetzner)
-3. **Configure environment variable:**
+3. **Configure Supabase** with pgvector extension
+4. **Run indexer** to populate knowledge base
+5. **Set environment variables:**
 
    ```env
    OLLAMA_URL=https://ai.yourdomain.com
+   NEXT_PUBLIC_SUPABASE_URL=your-supabase-url
+   NEXT_PUBLIC_SUPABASE_ANON_KEY=your-anon-key
    ```
 
 **Benefits:**
@@ -333,7 +372,11 @@ For production, run Ollama on a separate server:
 - ✅ Unlimited usage - no rate limits or quotas
 - ✅ Fast responses - no network latency (with good hardware)
 - ✅ Full control - choose any model, customize parameters
+- ✅ Platform-specific knowledge - AI trained on YOUR docs
+
+---
 
+<<<<<<< Updated upstream
 ---   ```bash
 
 # Add to Vercel environment variables
@@ -363,6 +406,8 @@ The AI tries providers in this order:
 
 ---
 
+=======
+>>>>>>> Stashed changes
 ## Feature Overview
 
 | Area | Summary |