On-prem AI
On-premises AI is AI that runs on hardware inside your own building instead of in a vendor’s cloud. The models, the agents and the data they handle stay on your network, under your control.
In a run
In the on-prem shape, every step of a run happens on your machine. The agent signs in, the access policy is checked, documents are read at the door, and the record is written. The record stays in your building with the rest of the job.
Our word for it
Runs in your building, on one small machine.
What runs on the machine
One small machine carries a whole Ares site: credentials for agents, access policies, inspection, workflows, the record and local models. Ares runs on any Kubernetes you put in your building. Nothing outside the building is needed to run it.
This matters when the data would hurt if it leaked: designs, pricing, customer files or controlled data. It also matters when a contract or a program rule says where processing may happen.
Where it stops
The models run on your machine: see local model for how they compare with cloud models. You provide the machine, and it sits on your network.
The same product in every shape
On-prem is one of three shapes. Cloud and hybrid run the same product, and credentials, access policies, inspection and the record work the same way in each. What changes is where they run. On-prem, sign-in comes from your own directory, and the record is kept in your building.
See the deployment shapes, or run the sandbox to see the same steps on demo data.
Related terms
- Air-gapped deploymentAn air-gapped AI deployment is a system that runs with no network link to the outside world.
- Local modelA local AI model is a model that runs on your own hardware rather than through a provider’s cloud service.
Tell us about one workflow.
We will say whether an agent should do it, and what the first step costs.
A senior member of the team replies within two business days.