Adaptive Responsive
Intelligent Assistant

Your personal AI assistant that works just like Jarvis. Fully customizable, privacy-focused, and built for the modern era.

Explore Architecture
~200ms
TTFA Response Time
4+
Platform Support
100%
Local Processing

Powerful Features

ARIA comes packed with advanced capabilities designed to enhance your productivity and daily life.

🎯

Dynamic MCP Tools

Support for stdio, SSE, and HTTP MCP tools with easy activation/deactivation through the WebUI. Self-activate based on context and needs.

🔊

Lightning Fast TTS

MiraTTS integration achieving ~200ms TTFA. OpenAI-compatible TTS servers also supported for maximum flexibility.

🎤

Advanced STT

Parakeet handles speech-to-text with high accuracy. Compatible with all OpenAI-compatible STT servers.

📷

Vision & Recognition

Connect cameras to satellites and use YOLO for people recognition. Train ARIA to recognize you and your loved ones.

🧠

Smart Memory

Long-term memory with RAG support. ARIA manages context, pruning old information and updating memories automatically.

⚡

Self-Activating

ARIA initiates conversations when needed. Detects new visitors, sets alarms, countdowns, and responds to your schedule.

Available Platforms

ARIA is built to work across all your devices with native performance.

🖥️

Desktop

Linux, Windows, and macOS

Built with Rust and .slint for native performance

📱

Mobile

iOS and Android

Native Flutter apps with fingerprint authentication

🏠

Satellite

Embedded Devices

Connect speaker and mic for full Alexa replacement

🌐

WebUI

Browser-based Control

Configure and use ARIA directly from your browser

Architecture

A robust, scalable design built for the future of AI assistants.

🎭 The Orchestrator

  • Headless server application
  • Controls all inputs/outputs
  • FastAPI-based core
  • WebUI for configuration
  • Direct usage interface

🔧 MCP Integration

  • Dynamic tool loading
  • Support: stdio, SSE, HTTP
  • Self-activate/deactivate tools
  • Tool discovery and management
  • Sandboxed execution

🧠 Built-in Tools

  • YOLO people recognition
  • Training pipeline
  • Camera integration
  • Alarm & countdown system
  • Schedule management

🗣️ Voice Engine

  • MiraTTS (~200ms TTFA)
  • Parakeet STT
  • OpenAI-compatible support
  • Streaming coming soon
  • Multiple voice options

🤖 LLM Support

  • OpenAI /v1/chat/completions
  • Anthropic Responses API (coming)
  • Recommended: GLM 4.7 Flash
  • Recommended: Qwen3.5 9B/35B
  • Multi-modal model support

💾 Memory & Context

  • Long-term memory storage
  • RAG implementation
  • Auto-pruning old data
  • Memory updates & learning
  • Custom WebUI settings

Privacy & Security

Your data, your control. ARIA is designed with privacy as a core principle.

No API Keys Without Consent

ARIA doesn't have access to any API keys unless you explicitly provide them.

Fully Local Processing

ARIA runs fully local unless you choose to use cloud models. Your data stays yours.

End-to-End Encryption

All communications between Applications and Satellites are encrypted and secured.

Clear Logging

Transparent logging system keeps you informed of all activities and decisions.

2FA & Passkey Support

WebUI supports two-factor authentication and passkey for enhanced security.

Fingerprint Authentication

Mobile apps support biometric authentication for quick and secure access.

Technology Stack

Built with cutting-edge technologies for optimal performance.

FastAPI Rust .slint Flutter MiraTTS Parakeet YOLO RAG OpenAI API GLM 4.7 Flash Qwen3.5 SSE HTTP stdio Kreuzberg Qdrant

Recommended Models: GLM 4.7 Flash, Qwen3.5 9B, or Qwen3.5 35B A3B for fast, nature-like answers and best tool-calling abilities.