ZenOS
Local AI Workstation
Run powerful AI agents directly on your Apple Silicon Mac. Zero cloud dependency, full privacy, and seamless integration with Claude Code.
Everything you need for local AI
Built for developers who want the power of cloud AI with the privacy and speed of local inference.
Multi-Model
Seamlessly route between Claude, GPT-4, Llama, and local models through a unified interface.
Local-First
All AI inference runs on your Apple Silicon. Your data never leaves your machine.
Vector Memory
Persistent semantic memory with Qdrant vector DB. Context that survives sessions.
Smart Routing
Intelligent model selection based on task complexity, cost, and latency requirements.
SSE Streaming
Real-time token streaming with Server-Sent Events. Low-latency responses you can feel.
Memory Consolidation
Automatically consolidate and compress conversation history for long-running sessions.
How it works
A clean pipeline from your AI agent to local models, with zero cloud dependency.
Up and running in minutes
Four simple steps to your own local AI workstation.
Install LM Studio
Download LM Studio and pull your preferred model (Llama 3, Mistral, etc.)
brew install lm-studioStart the Proxy
The ZenOS proxy starts on port 3001, ready to route requests
cd ZenOS/proxy && npm install && npm run devConfigure Claude Code
Point Claude Code at your local proxy instead of the cloud API
export ANTHROPIC_BASE_URL=http://localhost:3001
claudeStart Building
All requests now route through your local AI workstation
claude --model claude-sonnet-4-20250514Built with modern tools
A carefully selected stack optimized for performance on Apple Silicon.
Ready to go local?
Take control of your AI workflow. No cloud bills, no data leaving your machine, no compromises.