Onboarding a limited number of private deployments this quarter. Request access →
← Deployment

On-premises and isolated deployment

Air-gapped is a topology, not a checkbox.

An isolated Calliope AI deployment must account for infrastructure, images, dependencies, local models, identity, updates, evidence, and support without assuming an internet path. We scope that environment with the customer.

Start with what the network permits.

Some environments cannot use hosted model providers, remote control planes, package repositories, or automatic update channels. A credible design names those constraints before promising the same experience as a connected cloud deployment.

Calliope AI can be scoped for a customer-operated local environment, including local model endpoints and offline artifact transfer. The implementation is an architecture and services engagement, not the AWS self-service path with a different label.

Questions the deployment must answer:

The answers define the supported scope and operating responsibility.

  • Infrastructure · which Kubernetes or container baseline is available and who operates it
  • Inference · which local models and hardware can serve the approved workloads
  • Identity · how users and administrators authenticate without an external dependency
  • Artifacts · how images, packages, software bills of materials, and vulnerability data enter the environment
  • Updates · how patches are reviewed, transferred, tested, and rolled back
  • Evidence · where access records, runtime telemetry, and optional Zentinelle AI events are retained

Deployment and operations

01

Define the baseline

Validate the customer Kubernetes or container environment, storage, registry, identity, ingress, certificates, hardware, and operational ownership before installation.

02

Bring the complete artifact set

Transfer approved application images, dependencies, model artifacts, and integrity information through the customer's offline import process.

03

Operate without hidden callbacks

Use local model endpoints and customer-hosted platform dependencies. Document the update, support, telemetry, and evidence paths that remain available inside the isolated network.

Enterprises are scaling AI agents, data science, private LLMs and secure ML with Calliope AI

Self-host enterprise AI in days.

Stop choosing between moving fast and staying in control.

See how it works →