01
Operating responsibility
The order and architecture define which platform services Calliope AI operates, what customer administrators control, and how support and escalation work.
← Deployment
Managed Calliope AI Workbench
Use the Workbench without operating its infrastructure.
Calliope AI operates the Workbench environment, upgrades, and baseline platform services. Your organization still chooses the model endpoints, identity configuration, data sources, and workload policy appropriate for the deployment.
Choose managed when the missing piece is an operator.
A managed plan removes the day-to-day work of maintaining the Workbench baseline. It is useful for evaluation and for teams that want a supported environment without taking on platform operations.
Managed deployment is not the same trust topology as BYOC. Review the proposed account, tenancy, operator access, data sources, and model destinations with your security team before placing sensitive workloads in it.
What the managed Workbench includes:
The commercial Build environment plus a defined operating responsibility.
- Calliope AI Workbench · IDE, notebooks, chat, data tools, agents, and isolated workspace applications
- Managed baseline · infrastructure operation, upgrades, patching, backups, and platform monitoring within the purchased scope
- Identity and access · SSO, RBAC, and workspace access records
- Approved model access · bring provider credentials or select local endpoints appropriate for the workload
- Optional operating layers · Astrolift AI runtime and Zentinelle AI governance are separate products and services
What to review
02
Data and model flow
Document where workspace data is stored, which administrators can access the environment, and which hosted or local model endpoints may receive requests.
03
Expansion path
Move to BYOC when the workload requires a customer-owned data plane, or add Astrolift AI and Zentinelle AI when runtime and agent-level controls become part of the operating requirement.
