Where AI fits in a small property management company, and where it does not

Most of the job is messages, and most messages repeat

If you manage a few hundred doors with a small team, you already know the pattern. The same twenty questions come in every week, at all hours, through phone, text, email and the tenant portal. Answering them is not hard. Answering all of them, quickly, while also doing showings, inspections and owner statements, is.

That makes property management a good fit for AI in some places and a bad fit in others. The table is how I would split it.

Maintenance requests: triage first, then the work order

A maintenance request that says "the sink is broken" is not a work order. Someone has to find out which sink, whether it is leaking, whether water is getting into the unit below, and whether there is a photo. That back and forth is where a lot of a coordinator's day goes.

An AI assistant can do that intake by text or in your portal. It asks the questions your coordinator would ask, gets a photo, and sorts the request into emergency, urgent or routine based on rules you write. Routine requests become a clean work order in AppFolio, Buildium, Rent Manager or whatever you use, with the tenant's availability already attached. Some requests it can close without a vendor at all, like walking a tenant through resetting a tripped GFCI or a garbage disposal.

No heat in winter, no water, a gas smell, flooding, a fire, anything electrical that is sparking or smoking: those go straight to your on-call person. The assistant tells the tenant what to do right now (leave and call the gas company, shut off the water valve under the sink) and makes sure a human is on it. It does not decide whether a unit is habitable.

If you are on Rent Manager, I wrote specifically about Rent Manager integrations.

Leasing inquiries and showings, with Fair Housing in mind

Vacancy costs owners money every day, and a lot of leasing leads are lost simply because nobody replied for six hours. An assistant can answer the routine questions instantly (rent, deposit, pet policy, parking, laundry, move-in date), send the application link, and book a showing or a self-guided tour slot.

This is also where you need to be the most careful. The Fair Housing Act applies to every conversation with a prospect, including automated ones. Your assistant needs to give every prospect the same information, never steer people toward or away from a property, and never answer questions about who lives in the building or what the neighborhood is "like."

Move-in and move-out inspections from photos

A thorough inspection is dozens of photos, room by room, plus notes. Turning that into a report the tenant and owner can read is slow, so it often gets done thin, and thin move-in reports are why deposit disputes are hard to win.

AI can take the photos and a voice walk-through and produce a structured report: each room, each item, condition, and the matching photos. At move-out it can lay the move-in and move-out reports side by side so the differences are easy to see.

What it should not do is decide what the tenant owes. Normal wear and tear versus damage is a judgement call, and in Ohio the itemized list of deductions has to go to the tenant within 30 days after they move out and give a forwarding address. Use AI to get the evidence organized quickly. A person decides the deductions and signs the letter.

Owner updates that do not take a whole afternoon

Owners want to know what is going on, but most do not want to read a ledger. The monthly statement from your software is accurate and nearly unreadable to an owner with two duplexes.

AI can read the statement, the open work orders and the leasing activity and write a short summary for each owner: rent collected, what was spent and why, what is coming up, and how the vacancy is going ("14 inquiries, 5 showings, 2 applications this week"). You read it before it goes out. Owners get a clear picture and call you less.

This works well on top of whatever accounting your PM software already does. If any owners are on QuickBooks for their own books, QuickBooks automation covers moving the numbers across.

Vendors, keys and the coordination in between

A surprising amount of a coordinator's time is spent being the switchboard between tenants and vendors: who is coming, when, how they get in, whether the tenant needs to be home, and whether the job is actually done. Then the invoice arrives and someone has to match it to the work order.

AI can handle much of that relay. It confirms appointment windows with both sides, sends the tenant the vendor's name and time, asks the tenant afterward whether the problem is fixed, and matches vendor invoices to the right work order and property. Mismatches, like an invoice for twice the approved amount, get flagged for a person instead of paid. That last part is the same kind of work I describe on document processing.

Where AI stays out of property management

Property management carries more legal exposure than most small businesses, so I keep this list short and firm:

• Tenant screening decisions. A person, applying written criteria the same way every time.
• Legal notices. Late notices, notices to vacate, lease violations and anything eviction related follow your attorney's templates and state law.
• Deposit deductions. AI organizes the evidence. You decide the amounts.
• Habitability and emergencies. Always a person, always fast.
• Rent setting. Your call, with your owner.

Everything else on this page is about getting routine work done faster so that you have more time for those decisions. If you want help working out which piece to start with, how small businesses can start using AI is how I would approach it.