Home/Platform
How it runs

Your own J.A.R.V.I.S. On your own private server.

Every business gets its own private copy at its own web address, with its own core engine, its own data and its own AI key. We set it up, keep it updated and look after the server.

Platform
Platform

Private by design.

Private server

One business, one private copy.

Your address, your password, your data.

Each client runs in their own pair of containers on our server: the dashboard and the core engine. You sign in at your own address, such as jarvis.yourbusiness.com, behind your own password. Your data lives in your own volumes and survives every update.

  • Your own web address with HTTPS
  • One owner password, stored only as a salted hash
  • The engine is never exposed to the internet
Private server
Private copy per business
Core engine

The brain behind J.A.R.V.I.S.

Memory, skills, scheduled jobs and agents.

Each client has a private core engine that runs the long jobs: Tasks, Research, scheduled routines and missions. Anything it asks permission for waits in "Needs your OK". It cannot update itself and cannot be reached from outside your copy.

  • Skills and scheduled jobs
  • Background runs you can watch step by step
  • Pinned versions, tested before you get them
Core engine
Core engine
Online & offline

Full J.A.R.V.I.S online. A useful assistant offline.

Your data keeps building either way.

With a working AI key J.A.R.V.I.S is fully online. Without one it drops to "Offline · slow mode": mail sync and sorting, ranking, calendar, a daily brief built by code, figures, and short chat on a local model still work. Mail, calendar and notes keep building, so adding a key lights everything up with the full history.

  • Arc reactor dim offline, glowing online
  • No restart, no re-import when you go online
  • Offline mode arrives in 0.5
Online and offline
Online · offline · slow mode
Spend limit

Your key, your bill, your limit.

You pay the AI provider directly, with a monthly cap.

Each business uses its own OpenRouter key, so AI usage is billed to you at cost. You set a monthly spend limit in Settings; the arc reactor and orbs fill against it, with reminders at 80% and 100%. At the limit chat keeps working and background work pauses, with a "Raise my limit" button.

  • Monthly limit, resets on the 1st
  • Reminders at 80% and 100%
  • Limit arrives in 0.5; usage meters are live now
Spend limit
Spend against your limit
Approvals & safety

Nothing leaves the business without your yes.

Two presses for anything that goes to a customer.

Emails, posts, quotes and payments all wait for your approval in the app. Inboxes are listed without changing anything. There is no AI customer support yet: messaging apps are you talking to J.A.R.V.I.S only. And there is no bank access, ever.

  • Approval before anything is sent
  • Two presses for anything to a customer
  • No bank feeds or payout access, by design
Approvals
Approve, then it goes
Updates

New modules, without lifting a finger.

Every client runs the same core; modules are switched on per client.

New versions are released with plain-words notes and tried on our own demo first. Fixes go out straight away; new features wait for approval. Your settings, mail, notes and history are untouched by every update.

  • Plain-words release notes
  • Tested on the demo before clients
  • Your data survives every update
Updates
Release by release
Ready when you are

See the server side, not just the screens.

We walk through your address, the sign-in, the core engine and the approvals on a live copy.