Skip to main content

​
What is agtOS?

agtOS is an open-source, voice-native AI agent platform. It provides the infrastructure layer for AI agents that can hear, speak, think, and act — connecting voice input/output with LLM reasoning and tool execution.

​
Key Features

Voice Pipeline

STT → LLM → TTS cascade with WebSocket streaming, word timestamps, and barge-in support

Multi-Provider AI

Claude (API key + Max subscription), Ollama (local), model routing with automatic fallback

MCP Integration

9 built-in MCP tools + external MCP server connections for extensibility

Memory System

Three-tier: working (per-session), episodic (Redis), semantic (vector search)

Desktop App

Tauri 2 native shell with system tray, global hotkey, auto-update

Task Scheduling

Redis-backed cron, once, and interval task execution

​
Architecture

agtOS uses a dual-layer architecture:
  • Infrastructure Layer: Voice pipeline, audio processing, WebSocket transport
  • Orchestration Layer: AI-driven via protocol gateway (MCP + A2A ready)
User Voice → STT (speaches) → Model Router → Claude/Ollama → TTS → Speaker
                                    ↓
                              Agent Loop → Tool Registry → MCP Tools
                                    ↓
                           Memory (Working + Episodic + Semantic)

​
Getting Started

Quickstart

Get running in 5 minutes

Installation

Detailed setup guide