Understand the running system.
Bring services, model lanes, dependencies, and startup relationships into one operational inventory that shows how the environment fits together.
06 / Operations control plane
NbotaiOS
OSC.
Operate local AI infrastructure from one control plane. See services, models, routes, queues, health checks, capability tests, repair steps, recovery state, and execution evidence across the running NbotaiOS system.
Local AI services
Model routing + queues
Health + capability tests
Repair + recovery
Control grounded in
observable state.
OSC connects the system map, operational policies, health evidence, and recovery path. Operators can reason from what is actually running instead of working from disconnected assumptions.
Govern queues, policy, and approvals.
Keep work routes visible alongside queue state, approval points, and the policies that determine how an operation may proceed.
Test health and real capability.
Use audit events, health checks, and capability tests to distinguish a running process from a system that is genuinely ready to perform its work.
Repair from evidence.
Form repair plans from observed state and carry recovery through supervisor controls while keeping the route and result available for review.
Observe. Verify.
Recover.
The control loop begins with the real system, evaluates the route and capability, then applies a repair that can be checked against readiness evidence.
-
01 / OBSERVE
Read current state.
Map the service, model, dependency, startup, and queue state that defines the active operating environment.
-
02 / VERIFY
Test the route.
Check policy, approvals, health, and capability before deciding whether work should proceed or a repair is required.
-
03 / RECOVER
Restore continuity.
Execute the repair path through supervised recovery and confirm the resulting system state with fresh operational evidence.