Download the Nodofi AI Worker

Run an AI model on your own machine with Ollama and make it available over the relay — the planner behind Studio's Agentic AI, the Assistant, robot missions and your own Flows and services. No public address, no open port, your data stays yours.

Requires Node.js 18+ and Ollama

Windows, macOS & Linux

One package for every system (and Jetson / Raspberry Pi)

nodofi-ai-worker.zip

Unzip and run start.bat (Windows) or ./start.sh (macOS, Linux) — includes the install guide

Download .zip

Linux servers

The same package as a tarball

nodofi-ai-worker-linux-x64.tar.gz

For curl and tar on a headless machine

Download .tar.gz

Getting started

Install Node.js 18+ and Ollama, then pull a model: ollama pull gemma4.
Unzip the package and run start.bat (Windows) or ./start.sh (macOS, Linux). The first run installs what it needs and creates .env.
In .env, set RELAY_URL (the relay Studio uses) and WORKER_NAME — e.g. JozieAILLM2 for the Agentic AI planner, NodofiOllamaCoder for the Assistant. Run the start script again.
In Studio, point the AI Assistant (⚙ → Ollama) or the Agentic AI at the same relay and name.

Charge for it with MooChedda

Your GPU, your price. Set your MooChedda wallet and a price per request in .env; callers buy credit once with one USDC transfer to your wallet and each request uses a little of it. Requests that fail cost nothing, and every payment is checked with MooChedda before it counts.

MOOCHEDDA_WALLET=04…
CREDIT_COST_USDC=0.01
DEPOSIT_OPTIONS_USDC=5,10,25,100

See the AI Worker documentation for paying from Studio and apps, request signing, configuration and troubleshooting.