Deployment options
Choose a deployment based on where AI content may be processed and who will operate the service.
Glacis Hosted
Section titled “Glacis Hosted”With Glacis Hosted Proxy, Glacis operates the Proxy infrastructure. Your application sends supported requests to your assigned endpoint, and the Proxy calls your configured model provider when the controls allow it.
Your application can run outside Glacis infrastructure. A private connection changes the network access path; it does not turn a Glacis Hosted deployment into a self-hosted one.
Self-hosted
Section titled “Self-hosted”In a self-hosted deployment, your team operates the Proxy in an agreed environment. Confirm the supported installation, models, updates, evidence storage, support access and external dependencies with Glacis before choosing this option.
Self-hosted does not by itself mean that every component or evidence destination is local. The agreed deployment profile defines that boundary.
Confirm before connecting
Section titled “Confirm before connecting”| Decision | What to establish |
|---|---|
| Content processing | Where requests and responses are processed, including the selected model provider. |
| Service ownership | Who manages availability, updates, access, credentials and recovery. |
| Control coverage | Supported request types, controls and models for the workload. |
| Evidence | What is recorded, where it is stored, how long it is retained and how it can be retrieved. |
| Verification | Which verifier and trusted keys support the deployment’s receipts. |
Use the privacy and responsibilities guide when reviewing the data boundary. For installation and network configuration, use the private handbook for your deployment.