The only full-stack voice AI agent solution you'll ever need. Open-source firmware, AI agent, MCP servers, and AI models. Run everything privately on your own computers with zero cloud dependencies. Built with Rust for maximum performance and safety.
Proprietary AI systems from big tech are failing us!
Big Tech forces you to use their cloud services. They store your knowledge and even memory! You deserve the freedom to choose where your data goes.
Proprietary AI systems could be expensive, especially those requiring multimodal and real-time interactions. Bills add up fast.
Past generation's "smart speakers" or "smart assistants" like Alexa and Siri barely do anything useful. They're not actually smart.
Working with AI is a crucial job skill for the future. Yet, most existing AI are black boxes and impossible to learn. We need learnable AI systems.
Cheaper, better, and private realtime voice AI agent in your own house.
Speak with entertaining and engaging AI personalities right out of the box. No more boring, robotic responses.
Setup and run your own AI models on your computers without any monthly fees or usage limits. Complete freedom.
Ground your voice AI agent with your own documents, notes, and knowledge. Create a digital twin!
Make the AI agent speak in anyone's voice and style. Bring your favorite people from distance or past into your own house!
Control your environment and send commands across the Internet through natural voice interactions.
Build an end-to-end AI project and understand every component. Future-proof your job skills.
Complete full-stack open-source solution from hardware firmware to AI agents to MCP to AI models
Complete solution from embedded firmware to AI inference server. Run everything privately on your own hardware or use our hosted service - you decide.
Compatible with all major LLMs, voice-to-text, and TTS models. Choose your preferred AI models or switch between them seamlessly.
Create personalized voice experiences with advanced voice cloning technology. Make your AI sound exactly how you want.
Ground your LLMs with personal knowledge bases for accurate and updated responses that are aligned with your values.
Extensible action system with MCP-based tool calls. Enhance your AI's abilities with custom tools and integrations.
Use hosted services for convenience or run your own server for complete control. Unlike Big Tech, we give you the choice.
Built entirely in Rust for maximum safety, performance, and reliability. Memory-safe code with blazing-fast execution.
Full documentation and tutorials that teach you how to build and customize your own AI system on your own computers. You will learn AI from inside out.
Community-driven development. Learn how it works, fork it for your own needs, and contribute back!
WebSocket-based AI agent server with VAD, voice recognition and synthesis, LLM integration, and MCP support. The coordinator of your voice agent system.
Source code on GitHubHardware firmware written in embedded Rust for the EchoKit device. Handles audio I/O, buttons, display, WiFi, Bluetooth and device management.
Source code on GitHubVoice Activity Detection service using SOTA VAD models. Efficiently detects when the user is speaking for optimal interaction.
Source code on GitHubHigh-quality and streaming Text-to-Speech service with voice cloning capabilities. Generate natural-sounding speech in any voice you want.
Source code on GitHubFrom unboxing to chatting with your AI in minutes
Simply power up your EchoKit device and connect it to your WiFi network. No complex setup required.
Begin talking to your AI assistant immediately. It works out of the box with powerful default models.
Set up your own AI server, add custom knowledge, clone voices, and build advanced capabilities.