For many organisations the first question about an AI agent platform is not what it can do but where it runs. Customer data, internal documents and credentials often cannot leave the organisation's own cloud tenancy — sometimes not its own data centre. ATX Agent Hub is designed to be deployed where your data already lives.
Start with the free Starter plan or a guided evaluation to prove the use case. When it moves towards production, the Enterprise plan installs the platform inside your own environment.
The Enterprise plan ships the Terraform needed to provision ATX Agent Hub in your subscription, so the deployment is reviewable, repeatable and owned by your platform team. The application, its database, vector collections, run history and credential vault all live in your tenancy, under your network policies, your key management and your backup regime.
Model calls go to Azure AI Foundry or Google Vertex AI under your own cloud agreement — the same data-processing terms, region selection and quota you already negotiated — rather than through an additional AI vendor.
Some networks have no route to the public internet at all. For those, ATX Agent Hub can be deployed on-premise and air-gapped. The Admin Panel includes configuration for self-hosted models, so agents can use models running inside the same network.
Sovereign AI means the organisation keeps control of where its data is processed and which models touch it. Because model access is a per-workflow setting, a team can keep sensitive workflows on a model in its own region or network and use a different provider elsewhere — without rebuilding the agent.
Each project has DEV, TEST and PROD environments. Workflows are authored in DEV, tested, and promoted with independent approval, so the self-hosted platform fits the release process your organisation already runs for other production systems.
Yes. The Enterprise plan is self-hosted in your own Microsoft Azure or Google Cloud subscription, with Terraform included.
Yes. On-premise and air-gapped deployment is available on the Enterprise plan, including configuration for self-hosted models inside the network.
In your own tenancy: the application database, knowledge-base vector collections, run history and the encrypted credential vault all live in the subscription or data centre you deploy to.
Not in a self-hosted deployment. Model calls go from your deployment directly to the model endpoints you configure, such as your own Azure AI Foundry or Vertex AI.
This site needs JavaScript for the full experience. Every page above is reachable without it.