Entirely within your infrastructure.
Including the AI.
AnyAgent and its AI models can run on your own servers, under your security policies. Where that is not required, we offer to operate it for you.
- AI models on your serversNo model call leaves your network.
- Audit log stays with youEvery action is recorded on your own server.
- One outbound connectionIt carries the licence status, never your content.
Two decisions, any combination
Where AnyAgent runs and which AI models it uses are two separate questions. You can answer them once for the whole company, or differently for each process.
Who operates the AnyAgent server?
In your data centre or private cloud, under your security policies.
We can operate the server for you, at a location that suits your requirements, so you need no infrastructure of your own.
Which AI models does AnyAgent use?
- 1Open-weight models, run alongside AnyAgent
The model runs on the same infrastructure as the server.
- 2Open-weight models from an inference provider
Inside or outside the EU, often cheaper than your own GPUs.
- 3Commercial models from their provider
For example from Anthropic or OpenAI.
Different processes can use different models.
Where your data is processed
| Server operated by | Open-weight, run alongside | Inference provider | Model provider |
|---|---|---|---|
| Your company | Entirely in your network | Your network and the inference provider | Your network and the model provider |
| AnyAgent | AnyAgent | AnyAgent and the inference provider | AnyAgent and the model provider |
Fully on-prem, including the AI: your company operates the server and runs open-weight models alongside it.
What runs where
In on-prem operation, everything that touches your data runs inside your network. A single, narrow connection leads out: to our licence service.
Operates your applications on the workstation, like an employee would. Windows, macOS and Linux.
AI models, knowledge base, audit log and admin console. All process data stays here.
- AI models
- Knowledge base
- Audit log
- User accounts
- Admin console
Licences, device pairing and updates. Never receives your content.
This applies to on-prem operation with on-prem AI models. If AnyAgent is operated by us or uses cloud models, we agree the data flows with you in writing.
The licence check
AnyAgent checks its licence with our licence service at regular intervals. Only the licence status is exchanged, never your data. If the connection is interrupted, AnyAgent keeps working for several days, and your data remains accessible at all times.
- Regular checkLicence status only, no content
- Connection lostFor example a network outage
- Keeps workingFor several days, without restriction
- Your data stays accessibleAt all times, including export
Installation and operation
Container-based
Runs on your existing server infrastructure, in your data centre or private cloud.
Sized to your model choice
The hardware you need depends on the AI models you choose. We will work out the right sizing together with you before you invest.
Offline installation
An installation package is available for networks without direct internet access during setup.
Updates
New versions are provided through the licence service and installed in your environment.
Planning AnyAgent for your infrastructure
Together with your IT team, we review your requirements and propose a setup that suits your data and your budget.
Get in touch