A property manager running 200 to 800 units with three or four office staff spends most of the day on the same three things: reading tenant email, deciding which maintenance request matters right now, and hunting through leases for a clause somebody needs in the next ten minutes.

All three are text work. Text work is where current AI tools are actually good. None of it requires replacing your property management software.

Maintenance triage first

Maintenance is the best place to start because the cost of getting it wrong is visible and the volume is high. A mid-size office fields 40 to 150 requests a week, most of them written by a tenant at 11pm with no useful detail.

Point an assistant at the incoming request text and have it do three narrow jobs: classify the trade needed, flag anything that reads like water, gas, heat, smoke, or a lock failure, and write back the two questions a dispatcher would have asked anyway. Unit number, when it started, whether it is still happening.

The emergency flag is the part that pays. A tenant who writes "there is a wet patch under the sink and it smells funny" has filed a routine request in most systems and an urgent one in reality. Have the flag raise a human, never close a ticket on its own.

Pull last month's requests into a spreadsheet, run 50 of them through your assistant with that classification prompt, and compare its calls to what your team actually did. You will know inside an hour whether it is worth turning on.

Tenant email: draft, then send

The rule that keeps this safe is that AI writes drafts and a person sends them. Not because the writing is bad, but because tenant correspondence becomes evidence in a dispute and someone in your office needs to own every word of it.

The requests worth templating are the repetitive ones: rent reminders, notice of entry, parking and noise complaints, lease renewal outreach, move out instructions. Build ten prompts with your actual policies pasted in, so the tool answers from your rules rather than a general idea of how landlords behave.

Expect two to four minutes saved per message. At 60 tenant emails a week that is roughly three hours back, which is one afternoon a staff member spends on renewals instead of typing.

Set one rule in writing this week: nothing about late fees, notices to quit, or anything a court might read goes out without a named person reading it first.

Lease abstracts you never had time to build

Most small portfolios have leases in three or four formats going back a decade, and the answers live in PDFs nobody has read since signing. Current tools handle a 20 page lease well. Feed one in and ask for a fixed set of fields: term dates, rent and escalation, deposit, renewal and notice windows, pet and sublet terms, who pays which utility, and any rider that changes the base document.

Ask for a page or clause reference next to every field. That turns the output into something checkable rather than something you have to trust, and it means a disputed answer takes 30 seconds to verify instead of 20 minutes to find.

Have a person verify the first 20 abstracts against the source documents. If the reference format holds up over that sample, run the rest of the portfolio and keep the abstracts in a shared folder next to the leases.

Where to stop

Do not use AI to screen applicants, score tenants, decide renewals or non renewals, or write the substance of a notice. Fair housing exposure sits in exactly those decisions, and an automated output you cannot fully explain is a poor thing to defend. The same caution applies to marketing copy for listings, where wording about who a unit would suit is the wording that causes trouble.

Write those exclusions into a one page internal note so a new hire does not discover the boundary the hard way.

What it costs

A business chat account runs about $25 to $30 per user per month, and you need it for two or three people, not the whole office. That is roughly $75 a month to cover the front desk, the maintenance coordinator, and whoever handles leases. Business tiers matter here because the consumer versions default to using your conversations for training.

Your property management platform may already sell an AI add on. Check what it charges and what it actually does before buying anything separate, because a leasing chatbot bolted onto your existing system is often priced per unit and adds up faster than three seats.

A first month that proves it or kills it

Pick maintenance triage only. Run it for four weeks alongside your normal process, log the emergency flags it got right and wrong, and count hours saved on the inbox. If the numbers are thin, you have spent about $75 and learned something. If they hold, add lease abstracts next and leave tenant email for last.

If you would rather have someone set this up with you, the first call is free and takes about 30 minutes. We look at your request volume, your lease formats, and what your current software already includes, then tell you whether it is worth doing at all. Get in touch and pick a time.