AI for property management
Not a sales page. Four places AI genuinely works in a property management company today, two places you will be sold it and it does not work yet, and the one thing to check before you buy anything. From someone who audits operations for a living and sells no software.
No email required. No gate. From Ascend Systems · Charlotte, NC.
01
Four places. Each one names what it replaces, because if it does not replace a manual step you can point at, it is a demo.
01
An AI intake on the phone or by text takes the request when nobody can, captures the unit, what is happening and whether it is an emergency, sorts no-heat and active-leak from a dripping tap, creates the work order, and pages the on-call person only for the ones that need it. Anything with judgement in it goes to a person in the morning.
02
Given the tenant’s message and photos, an assistant drafts the work order in the vendor’s terms, picks the trade category, and writes the dispatch message. When the vendor’s invoice comes back, it reads it against the work order and flags the line that does not match. A person approves both.
03
An assistant over your own leases that answers “what does unit 4B’s lease say about pets” from the document, not from memory. The renewal letter drafted from the lease terms. The move-out inspection written from the photos. The vendor invoice read and checked before it becomes an owner charge.
04
The renewal notice ninety days out, from the lease end date already in the system. The owner update drafted from the month’s work orders and charges. The rent reminder before the late fee, not after. The message can be drafted by AI; the trigger is plain automation over dates and records you already hold.
02
Not never. Not yet, and not at the size of shop this page is written for.
Booking the tour and answering “is there parking” from the listing: real, and it works. Anything that touches who gets the unit is a decision with a law attached, and the vendor’s terms of service put that on you, not them. Use it to fill the calendar. Keep a person on every question that is really about an applicant.
Real for large portfolios with sensors in the mechanical rooms. For a few hundred units, the install dates already in your own system beat any model, and nobody has entered them. Put the dates in, set the reminder, and revisit the sensors when the portfolio is ten times the size.
A third, briefly: rent-pricing algorithms are a legal question before they are a technology question. Ask a lawyer before a vendor.
03
Before you buy any of the four above, trace one maintenance request through the company and count the re-types. A tenant text becomes a work order; the work order becomes a vendor dispatch; the vendor’s invoice becomes a charge on the owner statement; the statement lands in the books. In most companies each of those arrows is a person re-keying what the previous step already knew, and the vendor invoice is the one that goes through unread.
Every AI vendor sells you the front of that chain. If the finished work order still gets typed into the owner charge by hand, and the invoice still gets checked at month end from a stack, the AI made intake faster and the middle no better, and the office has a fourth place to look. Mapping the handoffs is the first thing an audit does, and it is the thing nobody selling software will do for you, because the answer is often “configure what you have.”
Two practical starting points: the platform you already pay for almost certainly has tenant-portal intake, vendor dispatch and renewal reminders you have not switched on; and the lease end dates and equipment install dates are the data that make the rest possible. Neither is AI. Both come first.
04
Which of the four above apply to you, which of the oversold ones you can safely ignore, and where the re-types actually are: that is the AI Audit. I map how a job moves through your shop, put a dollar figure on the hours that leak, and hand you a written blueprint that says what AI can take off the desk and what it cannot. Two weeks, sized by headcount.
A fixed fee, $5,000. If the recoverable savings in the blueprint do not cover it, you do not pay, and you keep the blueprint. The first step is a 30-minute call, and I will say on the call if it is not a fit.
05
The one already inside the platform you pay for, switched on and configured: tenant-portal intake, vendor dispatch, renewal reminders, owner reporting. Most companies have not turned them on. After that, a phone or text intake that can sort an emergency from a dripping tap, if after-hours requests are the leak. A standalone AI tool bought before either of those is a third place to type.
Intake and triage, yes: capture the unit and the problem, sort the emergency, create the work order, page the right person. Deciding what to do about it and which vendor to send is a person’s call, with the dispatch message drafted for them. The failure mode is not the AI getting it wrong; it is the AI creating a work order nobody reads.
The general-purpose assistants are free for reading a lease and drafting a letter, with one rule: nothing with a tenant’s name, unit or payment details goes in. The features inside your platform are free in the sense that you already pay for them. A “free” tool that wants your tenant data is not free.
Tour scheduling and answering questions from the listing, yes. Screening applicants and anything close to a fair-housing question, no. That is a decision with a law attached, the vendor’s terms put it on you, and a person should be making it.
Often nothing, because it is in the platform you already pay for. A managed intake service is a monthly fee. A build inside your tools is fixed-scope, $5,000 to $10,000. Finding out which of those you actually need, and which of the four above apply to your company, is the AI Audit: a fixed fee, $5,000, sized by headcount, not owed if the savings do not cover it.