Residents can tell. Whatever the vendor demo looked like, the person calling about that same repair (for the third time) knows within seconds whether the answering solution actually knows her building and recent call history.
In Gartner's survey this year, 87% of customers said that having the option to reach a person is essential when a company uses generative AI for service, and a separate poll of 6,000 consumers put the preference for a real person at 85% and rising. Those numbers are a warning sign about one specific way of using the technology, and unfortunately multifamily AI solutions are being sold that way at a lot of conference booths right now.
Meanwhile, the industry keeps framing the AI debate as adopt AI or don't, but the issue is more complex. There are two main ways to get AI wrong in multifamily, and we'll dive into both here.
The first trap: change nothing
The simplest option is to keep things the way they are right now. The phone rings, voicemail or an answering service takes it, and a person sorts it out later. From the office, this looks like the safe choice. To the resident, it mostly feels like silence; and to your on-call maintenance staff, it feels like less sleep.
This costs more than most operators realize. In our call data, about 35 to 40% of after-hours maintenance calls never needed a dispatch, and a technician got up for them anyway. A move-out runs nearly $4,000, according to Zego, whose latest resident experience research finds that controllable factors are often half or more of the reasons residents give for leaving. Doing nothing keeps the relationship “human” by default, but it leaves your team with work that never gets lighter and doesn’t always mean better service or relationships.
The second trap: automate everything
The other option, the one most demos are leading with, is to let the AI take over all of it. Anything from leasing questions to the collections call. That sounds appealing on paper, mostly because the demo is impressive.
A resident with a broken air conditioner makes her third call and reaches a bot that cheerfully asks her to describe the issue AGAIN. Gartner's own advice to service leaders is that generative AI shouldn't be a mandatory first step for every issue, and that people forced through multiple unsuccessful AI interactions are less likely to use the tool again. In an apartment community, a failed interaction costs more than a lost ticket, because the person on the phone lives in your building and decides once a year whether to renew.
Where we draw the line
We don't chase your residents for rent. We don't argue with them about repairs. That's a decision about the product, not a limitation of it — those are the conversations a community actually runs on, and letting software have those conversations is how a property ends up in the second trap while trying to avoid the first.
The line sits between the work and the relationship. Use AI for the work, not for the relationships.
What AI used for the work looks like
This is the easier half to define. Every resident call carries a set of questions that have correct answers. Is this an emergency? Which unit is it? What kind of repair? Is there already a work order open? What language does the caller speak? None of that takes judgment. It just has to happen quickly, often in the middle of the night, and usually without waking anyone up.
That's Lou's job. Lou is the AI that comes with HelloSpoke Assist, and it answers when your team can't, triages against your rules, opens the work order in Yardi, RealPage, Entrata, or ResMan, and follows up, in over 75 languages. For example, that 11:20 pm faucet call…Lou answers it, works out that a drip can wait, walks the resident in 31C through the shutoff valve, and files the work order overnight. The technician finds out at 8:00 am instead of at 11:20 pm and slept through the night. A real emergency still rings the on-call phone right away, with the details already gathered.
That's call-to-completion: the call gets answered, and the work it starts gets finished. Lou was trained on more than 100,000 real resident calls, and it resolves roughly 30% of conversations end-to-end with nobody on staff involved.
The relationship is the rest of this equation, and it's the part that takes judgment: a billing dispute, a decision about whether to waive a fee, or a resident who has now called three times about the same repair. On those calls, the most useful thing Lou can do is notice that and connect the resident with the right person.
For example, on Thursday's third call about the air conditioner, Lou checks Core and sees the work order on unit 8F was closed last week and that this is her third call about it. So it routes her straight to a person (with context) instead of making her go through it all again. That person picks up with the first request since they know it must be important. The resident finally reaches someone who knows the story. That's still from hello to handled, with a person doing the talking.
Both things can be true at once
Operators face two pressures at once. Adopt AI or fall behind, and don't put a machine between your team and the people who pay the rent. It’s sort of a trap: hold out and the work keeps growing, or automate everything and lose the residents you were trying to keep.
But HelloSpoke gives you an alternative solution that protects your relationships while automating work. Last night's after-hours calls mostly sort into two buckets: the ones that needed a person and the ones that just needed an answer about something less urgent. Our software handles the second kind and your team keeps the first.
FAQ
Should property management companies use AI for resident communication?
Yes, property management companies should use AI for the parts of resident communication that have a correct answer, such as triaging a maintenance call, opening the work order, or translating for a resident who's more comfortable in another language. The parts that need judgment should stay with a person, which is also what residents say they want.
What does "use AI for the work, not for the relationships" mean?
"Use AI for the work, not for the relationships" means AI handles the busywork around a resident conversation (answering, triage, logging, follow-up, and translation) while the actual conversation itself stays with a real person whenever it needs a judgment call. In HelloSpoke's case, Lou resolves about 30% of calls end-to-end and hands anything that needs a person to a person, with the full record attached.
Do residents actually mind talking to AI?
Some residents do mind talking to AI, and most want the option not to. In Gartner's 2026 survey, 87% of customers said it's essential to reach a human when a company uses generative AI for service. Mostly, residents mind being stuck more than they mind the technology itself.
Will Lou call residents about late rent?
No, Lou does not chase residents for rent, and HelloSpoke doesn't build collections into it. Rent conversations and repair disputes are relationship conversations, and those stay with your team.
See it on last week's calls
Book a demo and bring last week's calls. We'll show you which ones Lou would have finished on its own and which ones it would have handed to your team.
