Which maintenance calls should I give the assistant?
I begin with the address, unit, callback details, and the tenant's own description. For your Saugus management office, I want that record in the existing maintenance queue with a named staff owner. Receipt is not dispatch. I make the phone explain the next step without suggesting that a vendor is already coming. The tenant should not have to guess what the confirmation actually means.
My property management AI work starts with the process you already use. I ask how staff separate ordinary repairs from urgent reports and who covers the phone after the office closes. I need those instructions before I work on how the voice sounds. I do not present the assistant as an emergency response service. Its job is to pass the report into the right management process.
How do urgent reports get to a person?
I ask management to approve urgent-call wording with the appropriate professionals. If a report suggests immediate danger, the assistant needs to use that approved emergency direction and escalate to a person. I do not let it diagnose the hazard or ask a tenant to investigate it. I also check that the report cannot disappear into the ordinary queue. The urgent route has to change what happens next.
I need an owner on the other end of the on-call alert. Who receives it? How do they acknowledge it? Who takes over if they are unavailable? I test the backup with the assigned people. Sending an alert does not authorize the phone to say a technician is coming. I require a real acceptance from someone with dispatch authority before that message can go to the tenant.
What information belongs in the maintenance record?
I keep the tenant's words distinct from the assistant's summary. Water spreading stays a reported condition, not a machine's diagnosis. I also check the location carefully. Similar street names, separate buildings, and a missing unit can send the work to the wrong place. I have the assistant repeat the property and contact details so the tenant can correct them. The maintenance record needs to preserve those corrections.
For access, I collect the tenant's stated availability and send it to staff under your approved process. I leave entry rights and access disputes with people. Before setting recording, retention, or tenant-data rules, talk to your attorney for your situation. I keep access codes and unrelated tenant records out of routine alerts. Your staff need the maintenance request, without extra tenant information spreading through notifications.
How would I know the work order really arrived?
I place a test call, then open the maintenance system beside your coordinator. I check the property, reported issue, urgency flag, contact, and assigned queue. Next I interrupt the connection and look for a visible exception. The phone saying it received the request means nothing if staff cannot find it afterward. I need to see the work record where your coordinator actually manages repairs.
Then I call again about the same issue. I want staff to recognize a repeat without losing the new information. A worsening condition may need a different route. I also test a wrong address, a dropped call, and an unavailable on-call person. My article on decisions I would keep with people explains why I test those limits. The summary must not make the decision for management.
What would you set up for our management office?
For your office serving Saugus properties, I start with maintenance intake before adding other tenant questions. You and I define which records the assistant can read, which queue it can write to, and who handles exceptions. I leave vendor selection, spending approval, and tenant disputes with management. A working intake connection does not give the assistant authority over the rest of the property operation.
I finish the rehearsal with your actual on-call schedule and a written fallback for phone or software trouble. After launch, I compare call summaries with work orders staff used. Missing locations, repeat tenant calls, and alerts nobody acknowledged tell me where to correct the process. I want you to see who owns the request after the call ends. Another inbox is no help if everyone assumes someone else is watching it. The handoff has to be visible and usable.