Data.
Documents, records, prompts and answers stay in your environment and your country.
Some data cannot leave your control. SuperBrains can run inside your data centre or private cloud, including sites with no outbound connection, with open-weight models on your own hardware.
Documents, records, prompts and answers stay in your environment and your country.
Open-weight models run on your hardware. Outside providers are used only if you approve them.
Encryption keys and secrets are held by you.
Logs, audit records and updates follow your own processes.
| Option | Where it runs | Models | Typical use |
|---|---|---|---|
| SuperBrains Cloud | Managed by SuperBrains, in the region you choose | Approved providers under zero-retention terms | The fastest start |
| Dedicated | Single-tenant, managed by SuperBrains | Approved providers or open-weight models | Isolation without running it yourself |
| Your cloud | In your own cloud account | Providers or open-weight models | Data stays in your account |
| On-premises | In your data centre | Open-weight models on your hardware, or approved providers | Data stays on your premises |
| Air-gapped | In your data centre, with no outbound connection | Open-weight models and local search only | The most sensitive data, and plant networks |
Each use case gets the model it needs. Routine steps use smaller models, and harder reasoning uses larger ones. Models are chosen with your team against accuracy, speed, hardware and cost, and tested on your own examples before go-live. Search and embeddings run locally as well.
Agents, workflows, applications, the intelligence layer, guardrails and audit work the same way in SuperBrains Cloud and in an air-gapped site.
The forward-deployed team sizes the hardware, selects and evaluates the models, installs and hardens the platform, connects it to your identity provider and monitoring, and hands operations to your team with a support agreement.