How AI Routing Decides Urgent vs. Routine Maintenance

The agent does not decide on its own that a leak is bad enough to wake someone up. It compares the caller's answers, what happened, where, and whether it is active now, against the categories a property manager already wrote down and approved: active water intrusion, loss of heat, gas smell, unsecured access, and whatever else belongs on that specific property's list.

That separation matters. A written matrix produces the same routing decision on every call, no matter who is on shift or how calmly the caller describes it. A model improvising a severity judgment would not.

Three things drive the outcome: which category the reported issue falls under, whether the situation is active right now versus already resolved, and whether the caller described a safety risk. Those answers come from the same structured intake questions used to build the maintenance record, so classification and documentation happen from one call, not two.

A dripping faucet and a burst pipe can start the same way on the phone. The matrix, not the agent's impression of urgency, is what tells them apart.

A category match routes to on-call or the approved emergency path immediately, with the property, unit, and issue already attached so the person receiving it does not have to ask again. A confirmed non-match is logged with the same detail and queued for the morning team. A call that does not clearly fit either list, or that involves a safety concern outside the written categories, goes to the designated human review path rather than being forced into routine or urgent by default.

That third outcome exists on purpose. A rule set that only offers urgent or routine eventually gets stretched to cover a case it was never written for. Routing ambiguous calls to a person keeps the matrix honest.