15+ years in production
Imperal comes from more than fifteen years running real production systems — built from inside the problem it solves, not theorized from the outside.
About Imperal
Built from inside the problem.
Imperal is built by an operator who runs real production systems every day — not a research lab guessing at what operators need. The whole idea is honest architecture instead of industry exploitation: software that does the work, shows its receipts, and never acts behind your back.
One rule sits under everything here: you stay in control. That isn’t a tagline — it’s how the system is built.
The maker
Built by an operator, not a lab.
Most AI tools are designed by people who’ve never been on call for the thing they’re automating. Imperal is the opposite: it was built by someone who lives in the consequences.
Running a real business today
Imperal runs the day-to-day of a real multi-region hosting company right now — real customers, real money, real 3am incidents. Not a demo, not a waitlist.
What it is
An operating system for your work — not a chatbot bolted on.
Your computer has an operating system — the layer that takes what you want, decides what to do, carries it out, and keeps a record. Imperal is that layer for your work, run by talking instead of clicking. A chatbot bolted onto an app can answer questions, but it isn’t built to run your systems. Imperal puts the AI inside the operating system, so it can act — and the system, not the model, decides what’s allowed.
We call it the ICNLI AI Cloud OS — the first of its kind, built on an open standard, already running a real business.
The agent
Webbee is who you talk to. Say what you want in plain language; she works out the steps.
The kernel
The part that turns words into action it’s allowed to take: it checks what’s allowed, does the work, logs every step, and can undo. This is where “you stay in control” actually lives — the system decides, not the AI.
Extensions
New skills you can add — email, billing, servers, your own tools. The OS grows with whatever you run.
The rules it runs on are open — more on that below.
Open by default
An open standard, not a walled garden.
The protocol Imperal runs on is called ICNLI. We published it openly so it can outlive us — and so you never have to take our word for how it works.
Published in the open
The rulebook behind Imperal — ICNLI — is public and free to read, build on, or check, under a Creative Commons licence at icnli.org.
We wrote it, then gave it away
We published the standard openly and were first to run it end to end. Anyone can implement it — that’s the point of a standard.
No black box, no lock-in
Bring your own AI model. Run it on your own servers. Leave whenever you like. Nothing here depends on a secret you can’t inspect.
What we believe
Four things we won’t bend on.
You stay in control
Software should do what you ask and stop where you say. Anything that touches the real world waits for your explicit yes.
Honesty over hype
We say what the system actually does, show our receipts, and don’t dress a demo up as a product. Every claim should be checkable.
No lock-in
Built on an open standard, with your model and your servers if you want them. You can leave with your data whenever you like.
Your data is yours
One team’s data never leaks to another. Sensitive details are hidden from the AI by default. Nothing here is sold.
See it for yourself.
Connect your first tool and give it your first command — free, in a couple of minutes. You stay in control the whole way.
Want to talk first? Our contact details are in the footer below.