XRAI - Extensible Real-time AI Assistant

A modular, always-on AI voice assistant system with multiple implementations for different use cases.

Project Overview

XRAI is a collection of AI assistant implementations designed to provide always-listening, context-aware voice interaction with minimal resource usage. The project includes multiple approaches ranging from simple voice loops to advanced orchestrated agent systems.

Features

• Always-Listening Voice Interface: Multiple implementations for continuous voice interaction
• Modular Architecture: Choose from simple scripts to complex multi-agent systems
• Local-First Design: Run entirely on your machine without external dependencies
• API Integration Options: Support for OpenAI, Anthropic, and local models
• Memory & Context: Persistent conversation history and context awareness
• Desktop Integration: System overlays and HUD displays
• Developer Tools: Code analysis, GitHub integration, and predictive assistance

Components

Core Voice Assistants

• xrai.py: Main entry point and core functionality
• quick-voice-assistant.py: Simple 3-second recording loops with Ollama
• system-voice-agent.py: System-integrated voice control
• simple-always-listening-ai.sh: Continuous listening with VAD (Voice Activity Detection)

Desktop Overlays

• desktop-overlay.py: Desktop HUD for visual feedback
• system-overlay.py: System status and monitoring overlay
• geektool-style-overlay.py: Customizable GeekTool-style displays
• hud.py: Heads-up display for real-time information

Advanced Agents

• master-agent-orchestrator.py: Coordinates multiple specialized agents
• github-knowledge-agent.py: GitHub repository analysis and insights
• elite-repos-agent.py: Elite repository discovery and analysis
• predictive-agent.py: Predictive assistance based on context
• viral-innovation-agent.py: Innovation and trend analysis
• deep-code-analyzer.py: Advanced code analysis and understanding

Installation & Setup

• install.sh: Basic installation script
• install-complete-system.sh: Full system setup with all dependencies
• install-system-agent.sh: System agent specific installation
• auto-demo.sh: Automated demonstration of capabilities
• quick-preview.sh: Quick preview of functionality
• minimal-overlay.sh: Minimal overlay setup

Web Interface

• dashboard.html: Web-based dashboard for monitoring and control

Quick Start

Basic Voice Assistant

# Install dependencies
./install.sh

# Run simple voice assistant
python3 quick-voice-assistant.py

Full System

# Complete installation
./install-complete-system.sh

# Run main XRAI system
python3 xrai.py

Demo Mode

# See it in action
./auto-demo.sh

Requirements

• Python 3.8+
• macOS (primary support) or Linux
• Optional: Ollama for local LLM
• Optional: API keys for OpenAI/Anthropic

Architecture

The project supports multiple configurations:

1. Local-Only: Uses Ollama and local Whisper for complete offline operation
2. API-Based: Leverages OpenAI/Anthropic for higher quality responses
3. Hybrid: Falls back between API and local models based on availability
4. Multi-Agent: Orchestrates specialized agents for complex tasks

Memory & Persistence

• JSON-based conversation logs in ~/ai-logs/
• Daily logs with weekly summaries
• Context injection for continuity
• Optional Letta (MemGPT) integration for advanced memory

Current Status

• Voice Mode MCP Server integration complete
• Basic voice loops operational
• Desktop overlays functional
• Advanced agents in beta
• Memory system being refined

Future Roadmap

• Unified configuration system
• Plugin architecture for extensions
• Mobile companion app
• ESP32 hardware integration
• Advanced context understanding
• Multi-modal input support

Contributing

This project is under active development. Contributions welcome!

License

MIT License - See LICENSE file for details

---

XRAI: Your extensible, always-on AI companion