KaiROS AI

A powerful local AI assistant for Windows & Android
Run LLMs locally on your device • No cloud required • Privacy-first

---

📥 Download

• Download Latest Release - Windows Installer & Android APK
• Microsoft Store - Get it now
• Play Store - 🔜 Coming Soon!
• No .NET installation required
• Supports Windows 10/11 (x64) & Android 7.0+

---

🆚 Feature Comparison

Feature	Windows Desktop	Android Mobile
Local LLM Inference	✅	✅
Model Manager	✅	✅
Chat Interface	✅	✅
Chat History	✅	✅
System Prompt Editing	✅	✅
Custom Model Import	✅	✅
Markdown Rendering	✅	✅
RAG (Document Chat)	✅	✅
Local REST API	✅	❌
System Tray Support	✅	❌
DirectML & Vulkan	✅	❌

---

🖥️ Desktop Version (Windows)

The Desktop version is the full-featured powerhouse, designed for productivity and integration.

Key Features

• RAG (Retrieval Augmented Generation): Chat with your PDF, DOCX, and TXT files locally.
• Local REST API Server: Integrate your local models with VS Code (Continue), LM Studio, or your own apps.
• System Tray Integration: Keep your AI assistant running in the background.
• Advanced GPU Support: Full support for CUDA, DirectML, and Vulkan backends.

Desktop Screenshots

Model Catalog	Chat Interface

RAG (Document Chat)	Settings

---

📱 Mobile Version (Android)

The Mobile version brings the power of local AI to your pocket. Optimized for touch and on-the-go usage.

Key Features

• Offline Capable: Run LLMs anywhere, even without an internet connection (after model download).
• Battery Efficient: Optimized for mobile processors.
• Clean UI: A simplified interface focused on chat and quick interactions.
• Chat History: Save and resume your conversations anytime.

Mobile Screenshots

Chat Interface	Model Selection

Chat History	System Prompt

Settings

---

✨ Shared Features

Core Capabilities

• 🤖 Run LLMs Locally - No internet required after model download
• 📦 Model Catalog - 31 pre-configured models from 9 organizations
• ⬇️ Download Manager - Pause, resume, and manage model downloads
• 💬 Streaming Responses - Real-time text generation
• 📊 Performance Stats - Real-time tokens/sec and memory usage

Model Catalog

• 🏢 Organization Sections - Collapsible groups for Qwen, Google, Meta, Microsoft, and more
• 🔍 Advanced Filtering - Filter by Organization, Family, Variant (CPU-Only, GPU-Recommended)
• 🏷️ Visual Badges - Category, family, variant, and download status indicators
• ➕ Custom Models - Add your own GGUF models from local files or URLs

Advanced

• 🎨 Modern Dark Theme - Beautiful gradient-based UI design
• 💬 Feedback Hub - Send feedback directly from Settings

---

🔌 Local REST API (Desktop Only)

Build AI-powered applications without cloud dependencies!

KaiROS AI includes a fully local REST API server - perfect for developers who want to integrate local LLMs into their applications.

Quick Start

# Check status
curl http://localhost:5000/health

# Chat (non-streaming)
curl -X POST http://localhost:5000/chat \
  -H "Content-Type: application/json" \
  -d '{"messages":[{"role":"user","content":"Hello!"}]}'

enable in Settings → API Server

---

🚀 Getting Started

Prerequisites

• Windows 10/11 (x64)
• Android 7.0+ (API 24+)
• .NET 9 SDK - Download
• CUDA Toolkit 12 (optional, for GPU acceleration) - Download

Installation

1. Clone the repository

   git clone https://github.com/yourusername/KaiROS.AI.git
   cd KaiROS.AI

2. Restore packages and build

   dotnet restore
   dotnet build --configuration Release

3. Run the application

   dotnet run --project KaiROS.AI

📦 Model Catalog Overview

Supported Organizations

Organization	Highlights
Qwen	Qwen 2.5/3 series (0.5B - 14B) - Excellent multilingual
Google	Gemma 2/3 models (270M - 27B) - High quality
Meta	LLaMA 3.1/3.2 + TinyLlama
Microsoft	Phi-2, Phi-3, BitNet b1.58
MistralAI	Mistral 7B, Mistral Small 24B
Open Source	GPT-oss 20B ⚠️ Experimental

Recommended Models ⭐

• Phi-3 Mini 3.8B - Best for general conversations (4 GB RAM)
• Qwen 2.5 3B - Excellent multilingual and coding (4 GB RAM)
• Mistral 7B - Complex reasoning tasks (8 GB RAM)

---

🛠️ Tech Stack

• Framework: .NET 9 + WPF (Windows) / MAUI (Android)
• LLM Runtime: LLamaSharp 0.25.0
• MVVM: CommunityToolkit.Mvvm 8.4.0
• GPU Support: CUDA 12, DirectML, Vulkan
• Model Format: GGUF (llama.cpp compatible)
• Database: SQLite (for custom models)

📁 Project Structure

KaiROS.AI/
├── Assets/              # App icons and images
├── Converters/          # XAML value converters
├── Models/              # Data models
├── Services/            # Business logic
├── Themes/              # UI styling
├── ViewModels/          # MVVM ViewModels
├── Views/               # XAML views
└── appsettings.json     # Model catalog config

🤝 Contributing & License

Contributions are welcome! Please feel free to submit a Pull Request. This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• LLamaSharp - Excellent .NET bindings for llama.cpp - This project wouldn't be possible without LLamaSharp!
• llama.cpp - High-performance LLM inference in C/C++
• Hugging Face - Model hosting and community

---

Made with ❤️ for local AI enthusiasts