AI leasing covers a range of things sold under one phrase, from answering enquiries at midnight to scoring applicants. The first end is useful and low risk. The far end touches decisions about who gets housing, and that is regulated ground where the tool's convenience is not the relevant consideration. Knowing which end a product sits at is the whole of the buying decision.
What it does well: the response gap
Most leasing loss happens in the hours between an enquiry and a reply. An assistant that answers immediately, answers accurately about availability and price, and books a viewing into a real calendar, closes that gap without making any decision about anybody. This is the majority of the measurable benefit and it carries almost none of the risk, which is why it is where most portfolios should start and stop.
Scheduling, tours and follow-up
Self-guided tours, access codes, reminders and post-viewing follow-up are scheduling problems, and automation handles them reliably. The thing to check is the fallback: what happens when somebody asks a question the assistant cannot answer. A tool that guesses is worse than one that hands over, because a wrong answer about a pet policy or a fee becomes something you said.
Where it must stop: decisions about applicants
Screening, scoring and any automated recommendation to accept or decline touches federal fair housing law, which prohibits discrimination in the terms and conditions of a rental and does not care whether the discrimination came from a person or a model. A system that produces a disparate outcome is the landlord's problem, not the vendor's. If a product scores applicants, you need to know what it scores on and be able to explain any decline in terms a person can check.
What to ask a vendor
What data the model was trained or prompted on. Whether it ever declines or deprioritises an applicant, and on what basis. Whether every conversation is retrievable in full, because your fair housing defence is the transcript. And whether it can be configured to answer only from your own stated policies, which is the setting that turns a plausible answering machine into a reliable one.
Questions people ask about ai leasing
Does an AI leasing assistant need to identify itself?
Several jurisdictions now require disclosure when a consumer is talking to an automated system, and it is good practice regardless. Tenants find out anyway.
Is automated screening prohibited?
Not as such, but the outcome is judged the same way a human decision is. The risk is not the automation, it is being unable to explain a decline.
What is the realistic gain?
Faster first response and fewer missed enquiries. Those move occupancy more than anything else on the list, and neither requires the tool to decide anything.