- What’s the difference between self-host and hosted?
- Self-host runs the daemon on your own machine: free, source-available, and yours to manage, with your own model keys. On a paid plan we run it for you, we keep it updated and backed up, and Jarvis AI is included. It’s the same agent OS either way.
- Do I need my own model API keys?
- Not on any paid plan. Every paid plan is hosted with the AI included, so there are no keys to configure and no model bills to reconcile. Self-host is the only option where you bring your own keys, and it’s free.
- What happens when I reach the limits?
- It degrades, it doesn’t block. There is one allowance and everything draws from it, typing and talking alike. Past it, work keeps running on a smaller set of models for the rest of the month, voice included. There’s no overage billing at launch, so nothing arrives on your card that you didn’t choose.
- What runs on my machine versus your servers?
- The brain, LLM routing, the knowledge Vault, and the workflow engine, is hosted. The sidecar and the Pebble that watches your screen and acts for you stay on your device, connected over an encrypted WebSocket. One brain, your machines.
- Can I switch plans or cancel?
- Anytime, from Settings · Billing. Plans are monthly and switching is prorated. Cancelling stops the next renewal and keeps your access until the period ends.
- When do the paid plans open?
- They’re rolling out now. Join the waitlist and we’ll email you the moment your plan is available. Self-host is ready today, no waiting.