58% of the real estate management professionals who answered a 2026 survey by the Institute of Real Estate Management (IREM) and the software company AppFolio said they use AI, up from 21% in 2023. Getting access to AI isn't the hard part anymore. Deciding what it's allowed to do is.

How can you use AI in property management?

Two-thirds of an onsite week goes to routine and reactive work. A 2025 survey by the National Apartment Association (NAA), run with AppFolio, found onsite professionals spend 42% of the week on routine operations, paperwork and data entry, and another 24% on reactive work like urgent requests. A lot of that is reading, pulling things together and writing a first draft. An AI teammate can take that part.

Where an onsite property team's week goes

Share of the work week, as onsite professionals reported it in the NAA survey

National Apartment Association and AppFolio, 2025 Performance Ecosystem survey.

Job	What the AI reads	What you get back	Who should approve
Monthly owner report	Property system reports, the books, last month's report	A drafted report, with the source for every number	The property manager, before it goes to the owner
Maintenance requests	Emails, texts and portal requests, plus the unit and vendor history	Requests matched to the unit and vendor, with a drafted reply and work order	The maintenance coordinator, before any work order or spend
Owner and tenant replies	The ledger, work orders, the lease and past threads	A drafted reply with the evidence attached	Whoever looks after that owner or tenant, before it's sent
Late rent and renewal reminders	The rent roll, payment history and lease dates	One drafted message per resident, based on what they owe and when their lease ends	The property manager, one approval per message

Compile the monthly owner report

The numbers already exist. Somebody pulls them out of the accounting tool, checks them against the property system, writes the same three paragraphs of commentary they wrote last month, and sends it to each owner. Every month, the same route through the same tools.

Ask for it in the channel where your team already talks and the report comes back written. You read it, you fix the one line you disagree with, you approve the send.

See it run Pull the numbers an owner report starts from One question returns occupancy, open work and unpaid rent for every property in a region, with the filter shown so you can check it.

Triage maintenance requests

Requests arrive by email, by text forwarded into chat, and through the portal. Someone reads each one, works out whether it is an emergency, matches it to a property and a vendor, and writes back.

The reading and sorting part is mechanical. Let the AI read what comes in, pull the property and tenant history, flag what looks urgent, and put a drafted reply and work order in front of a person.

When the person on call phones a resident about an emergency, the unit's history is already in the channel. And someone still reads the whole inbox at set times, not only the flagged items, so a burst pipe filed as routine still gets seen.

See it run Route a maintenance approval before spend Dash prices a repair against the same job last year, routes the approval card, and holds the work order until someone answers.

If requests still come in on paper, our free maintenance request form has a box for the date you received each one.

Draft the owner and tenant replies

An owner asks why the repair cost what it did. A tenant asks about a lease clause. The answer is sitting in four places: the ledger, the work order, the lease, and a thread from March.

Gathering that is the slow part. Writing the reply once you have it takes two minutes. Hand the gathering to AI and you start from a draft, not a blank reply.

See it run Pull a resident's payment history before a hard conversation One question returns what the resident owes now, how they've paid over two years and when the pattern changed.

Chase the follow-ups nobody owns

Renewals coming up in sixty days. Insurance certificates that expired. Turnovers where the third step stalled and no one noticed. Deposits not yet returned.

This is the work that quietly costs money, because it only becomes visible when it has already gone wrong. Ask for the same check every Monday morning and get one list in your channel, with the drafts ready for the ones that need a nudge.

See it run Draft rent arrears follow-ups for approval Dash drafts a follow-up for each resident behind on rent and holds every one in the thread until a person approves.

Each demo is a real conversation you can read end to end. See all property management demos.

Which property management decisions should AI not make?

We'd keep three with a person. The first two have already ended up in court.

Who gets the apartment

The Fair Housing Act covers screening whether a person or software makes the call. In May 2024, the U.S. Department of Housing and Urban Development (HUD) said the Act applies to tenant screening and housing ads, "including when artificial intelligence and algorithms are used to perform these functions." HUD withdrew its guidance on AI in housing ads in September 2025, but that doesn't change the Act.

In 2024, a federal court in Massachusetts approved a settlement of up to $2.275 million in a lawsuit against the tenant-screening company SafeRent. The suit alleged its screening score discriminated against Black and Hispanic applicants with housing vouchers. SafeRent agreed to stop giving approve-or-decline recommendations for voucher applicants unless civil rights experts validate its model.

So our rule for a small office is short: don't paste applications into a chatbot and ask who to approve. Let AI collect the documents and flag what's missing. A person makes the call.

What the rent should be

The Justice Department sued RealPage in August 2024, saying its rent-setting software used landlords' nonpublic data to set prices, and filed a proposed settlement in November 2025. Have AI pull comparable rents from public listings and your own rent roll, never from data other landlords share privately. Then set the number yourself.

What a resident hears about an urgent repair

For everyday questions, residents want speed: Buildium's 2026 Renters' Survey found 84% of renters expect a reply from their property manager within 24 hours. AI drafting helps you hit that. For something urgent, they want a person. In a 2025 survey of apartment residents commissioned by the maintenance software company HappyCo, 82% wanted a human first for an urgent maintenance issue.

So keep emergencies on a phone line a person answers, day and night. We've written up where resident chatbots go wrong.

Should you use your property software's AI, a chatbot or an AI teammate?

An AI teammate, for any job that crosses tools and people. In a property office, that's most of them: the owner's email, the vendor invoice in the shared inbox, the spreadsheet someone keeps on the side, and the colleague who has to approve the send.

AI built into your property system works on the data in that one system. A general chatbot works on what one person pastes into it, and that person copies the answer out and sends it by hand. Neither sees the whole job.

We built Dash for the whole job. Dash is an AI teammate that works with your team in Microsoft Teams, Slack, text and web chat. It connects to Buildium, Rentvine and DoorLoop, reads your AppFolio reports and reaches 3,000+ other tools. For a tool it can't connect to directly, it works through the browser.

Dash looks things up without asking. By default, anything that sends, posts, writes to a system or spends money waits in the thread for an approval card everyone in it can see. Anyone with access to that tool in Dash can approve. An admin can limit a tool to named approvers, like your maintenance coordinator for work orders.

Weighing tools for your office? See Dash's AI property management software, or compare AI agents for property management.

Is AI going to take over property management?

No. It takes the busywork, not the job. It won't know which owner needs a phone call rather than an email, and it shouldn't decide whether to evict or whether to cover a repair cost to keep a good tenant. Those calls stay with your team.

How do you start using AI in a property management office?

Do not start with the most impressive thing you can think of. Start with the task your office repeats every single week and nobody enjoys.

Pick one recurring job. The Monday roll-up, the owner report, the maintenance inbox. One, not five.
Write down what needs a person. In the IREM and AppFolio survey, 73% of respondents said they check AI output before acting on it, but only 8% of organizations had a written AI policy. One page that lists what AI may draft, what a person approves and what it never touches is enough to start.
Run it beside the person who does it now. Compare the draft against what they would have written.
Check that you can inspect the work. You should be able to see where every number came from.
Check what happens when it is missing something. It should stop and ask, not guess.

If the draft is good enough that the person is editing rather than rewriting, you have your answer. Then add the second job.

Dash is per workspace rather than per seat, there are no contracts, and new workspaces start with free credits and no credit card.

Add Dash to your team

AI for property management: what to hand off and what to keep