What is Resident AI? AI that lives at your business.
Resident AI is artificial intelligence that lives at your business. The model, your documents and the index run on hardware inside your own premises, you hold the keys, and a provider keeps it running. Nothing leaves the building unless you open a door.
Most businesses meet AI through one of three routes. Frontier cloud services sell the most capable models to everyone. Sovereign inference services sell similar models with an Australian address. GPU clouds rent dedicated hardware in a data centre for you to build on. All three sell computing to people who then build something, or who are comfortable with their work passing through someone else’s building.
A professional practice usually wants something else: a finished system, at its own premises, that someone else maintains. Judged on token prices, benchmark scores and GPU hours, that system looks like none of the three. Resident AI is the name for it, so it can be judged on the questions a practice actually asks: where the work goes, who holds the keys, and who keeps it running.
The residence test
Three conditions, and all three are required. A system that fails one belongs in a different category.
The model runs on hardware physically at your business. Switch off the internet and it still answers.
You control the admin, who has access to what, and whether anything connects to it at all.
Someone else installs, updates and monitors it. Nobody at your business ever needs to open a terminal.
The third condition is what separates a category from a parts list. Without it you have self-hosted AI, which suits businesses with their own IT team and very few others.
Two questions that sort every option
If you pulled the internet cable, would your AI still work? Only options running on your own premises answer yes.
And who fixes it when it breaks? Of those, only Resident AI answers with someone other than you.
The map, by who holds the keys
Custody is the axis: who holds the keys to the machine your work passes through.
| Category | Where the model runs | Who holds the keys | Typical examples |
|---|---|---|---|
| Tenant AI | Shared cloud infrastructure, anywhere | The vendor | ChatGPT, Microsoft Copilot, Gemini |
| Onshore AI | Shared infrastructure in Australia | The vendor | Sovereign inference services |
| Dedicated cloud AI | A machine rented in a data centre | The operator | GPU cloud providers |
| Self-hosted AI | Your premises | You, and you also run it | A server from a hardware vendor, or a build of your own |
| Resident AI | Your premises | You, with a provider running it | Ermos Dominion |
Tenant AI is the right choice for a great deal of work. Every major cloud model is multi-tenant: many customers’ prompts are processed on shared infrastructure the vendor controls, and the reputable vendors protect it well. What changes along the map is custody. Our on-premise vs cloud comparison and private cloud comparison go row by row.
Residence and data residency
The two words sound alike and answer different questions. Data residency keeps your data in the country. Residence keeps your intelligence in the building. An onshore cloud service can provide data residency; only a system at your own premises can provide residence.
Who Resident AI suits
Professional practices whose obligations to clients call for custody of client work: law, accounting, medical, financial advice and broking. Practices large enough to use AI every day across the whole team, and too small to run an IT department. For most of them, one unit serves everyone.
Large organisations with the same constraint usually have IT teams that build or rent their own infrastructure. Businesses without the constraint are well served by tenant AI, or by an onshore service such as Ermos Edge.
Where Ermos fits
Ermos Dominion is Australia’s first Resident AI for professional practices. An Ermos AI unit sits in your office, and the model, your documents and the index all run on it. Nothing connects to it unless you allow it, and it keeps answering with the internet switched off. Ermos installs, updates and monitors it: support happens through sessions you start and supervise, and monitoring watches the hardware, never your files. ErmosIQ runs only on Dominion, because it needs the model on your own hardware. The unit adapts to your work over time, and nothing your team teaches it is reused elsewhere.
Ermos Edge is the rented option. The same software, hosted for your business in Australia, with answers generated by a named Australian provider. Your documents stay on your own instance, and only the excerpts a person is permitted to see go, with their question, to that provider. Some businesses run Edge for years. Others start there and move to Dominion later, and their data moves with them.
Common questions
What is Resident AI?
How is Resident AI different from on-premise AI?
Is Resident AI the same as data residency?
Does Resident AI need an internet connection?
Is Ermos Dominion Resident AI?
See Dominion answering from your own documents.
A demonstration takes 30 minutes, uses no client data, and comes with no obligation.
Book a demo