STRATUM AI Overview
STRATUM AI is an operational layer for infrastructure and guest work. It is separate from GPU acceleration and expands through Skills, content, node-level operations, and AI Studio.
Operational AI
STRATUM AI works against infrastructure and guest context that already exists in STRATUM. A node-level task can be scoped to a selected system, while Skills and controlled content provide reusable operational knowledge.
Sovereign AI for datacenter operations
Sovereign AI is an operating model in which the organization retains authority over infrastructure, workload location, model access, operational data, and what automation is allowed to change. STRATUM AI separates model reasoning from the STRATUM execution harness: the model can propose an operation, but STRATUM enforces the configured policies, approval boundaries, Skills, mission scope, and verification.
Choose approved model providers and deployment boundaries according to your data residency and security requirements. STRATUM does not make an external model provider sovereign by itself; sovereignty also depends on the selected provider, network policy, identity controls, and where information is processed and retained.
For governance and traceability, see Guardrails & Operator Behavior and AI TRACE.
Separate from GPU
GPU & Guest Tools describes how applications consume accelerator resources. STRATUM AI describes how operators use AI to reason about and act on the datacenter. An AI workflow may use accelerated models, but the product concepts are independent.
Scheduled and recurring AI operations
AI Schedule adds time as another input to the STRATUM AI operating model. Create one-time, hourly, daily or weekday-based recurring operations against a captured infrastructure focus; when the schedule fires, the request becomes a normal governed STRATUM AI Mission instead of an untracked background script.
AI Studio
AI Studio is the DAG-based composition environment for the STRATUM AI Operator. Administrators assemble specialist Roles, Skills, live context, provider connections, authority, guardrails, compliance overlays, behavior, approval, execution and verification as a typed directed acyclic graph.
The DAG is not just a diagram. Typed edges define how intent, capability, policy, authority, evidence and execution state move forward through the Operator. Start with AI Studio Concepts & Operator DAG before building or modifying an Operator.
