Neural Metropolis OS
Intelligence, within boundaries.
Human oversight. Enforced limits. Your metropolis operates on your terms.
Every automatic action passes the safety layer
Only green, reversible decisions within policy limits can execute automatically.
Safety distribution
Classification of the latest 0 simulated decisions
Autonomous & reversible
Low risk · safety-validated
Controlled testing
Medium risk · limited scope
Explicit human approval
High risk · significant impact
Policy guardrails
Maximum automatic traffic shift
Per autonomous decision
Automatic change budget
Rolling 60-minute window
Maximum execution attempts
Hard ceiling: 3 attempts
Policy limits apply to automatic actions. Orange and red decisions require explicit operator approval. The kill switch blocks automatic and approved execution.
Safety event log
Loading safety events…