Getting started

Mandala is self-hosted: it runs on a machine you control, and your other devices join it. Getting started is four decisions and about half an hour.

1. Where it runs. A Mac, a home server or a VPS. See Choose where Mandala runs.
2. Your account. A first name and Face ID. See The first run.
3. What powers your accomplice. A ChatGPT or Grok sign-in, an API key, or a model on your own machine. See Choose a model.
4. Who your accomplice is. From a plain assistant to an accomplice with a personality. See Meet your accomplice.

In this part

• What is Mandala
• How Mandala fits together
• Choose where Mandala runs
• Install on Linux
• Install on a Mac
• Install on a VPS
• The first run
• Choose a model
• Meet your accomplice
• Your first hour

What you need

• A Linux machine you control (a home server, a mini PC or a VPS), or a Mac with Apple silicon.
• An AI model: an OpenAI, Anthropic, xAI or other API key, a Grok sign-in, or Ollama running locally.
• Tailscale on the server and on every device you will use Mandala from: it gives Mandala a private HTTPS address only your devices reach. See Tailscale and HTTPS.
• A device with Face ID, Touch ID or Windows Hello (or a phone to sign with) for your account.