Tenant Screening Software and AI: How Far Is Too Far?

Can AI pick your tenants for you?

One company that tried at scale is SafeRent, which ran an algorithm scoring rental applicants for landlords. The outcome was a 2.28 million dollar settlement approved by a U.S. federal court in late 2024, along with the company agreeing to stop producing automated accept/reject recommendations for applicants using housing vouchers. The lawsuit alleged the algorithm disproportionately screened out Black and Hispanic applicants in that group. So the honest answer today is: yes, it can pick, and you should not let it.

Does that mean a landlord with several units, or a property management office, should stay away from automation? Quite the opposite. Most of rent management (payment reminders, collection records, lease and increase calendars) is exactly the kind of work software does well. This article draws the line: which jobs can rent collection software take over with a clear conscience, and where should tenant screening stop? For the bigger picture of which AI uses fit which industry, our industry-by-industry AI guide is a good companion.

What does rent collection software actually take over?

Rent collection and property management software gathers your units, leases, and payments on one screen, reminds tenants automatically before rent day, records payments, flags late ones, and puts lease renewals and rent increases on a calendar. It does what your spreadsheet does, without requiring you to remember anything.

The market is crowded and, for small landlords, surprisingly cheap. Platforms like Buildium, TurboTenant, Avail, Innago, and Azibo compete hard for the independent-landlord segment; several of them charge landlords little or nothing and pass the screening fee to the applicant instead. Market researchers disagree wildly on how big the property management software market is (estimates for 2025 range from under 4 billion to over 26 billion dollars depending on scope), but they agree on the direction: steady growth for years to come.

The question to ask is not price but coverage: does it send reminders, handle partial payments, report arrears, and encode the rent increase rules that actually apply where your property sits? That last item deserves its own section.

Is a spreadsheet enough?

It is; up to a point. Two or three units, reliable tenants, and a well-kept spreadsheet can coexist for a long time. The spreadsheet's limit is not calculation but recall: it will not tell you a due date on its own, will not flag an upcoming renewal, will not message a tenant.

The breaking point we see in practice is around five or six leases. Past that threshold, tabs multiply, versions fork (you know the file called "rent_tracking_FINAL_v3.xlsx"), and preparing an owner report becomes a chore in itself. Then there is collaboration: once two people in an office update the same sheet, an error is only a matter of time. The reason to switch is not that spreadsheets are bad; it is that the portfolio has outgrown the job spreadsheets were designed for.

Rent increase rules are local: let software do the math, then verify

In many rental markets the annual increase is not a free choice: jurisdictions from Germany to Turkey to several U.S. states and cities cap increases, often tying them to an official inflation index, with the applicable month and formula fixed by law. Good software encodes the local rule, warns you as a renewal approaches, applies the current index, and produces the new figure.

Getting this wrong costs money in both directions. An increase you forget to apply is straight lost income, and in indexed systems it also lowers the base for every following year, so the loss compounds. An increase above the legal cap can be void, and in some jurisdictions tenants can reclaim overpaid rent years later. Portfolios run by memory, or by a blanket "same raise for everyone in January" habit, reliably produce both errors.

One caution: because the applicable index month usually depends on each lease's renewal date, check the software's math against the official published index for your first few renewals. Trust the automation, but verify it once before you do.

How far should AI go in tenant screening?

AI-assisted tenant screening is a mature industry, especially in the U.S.: credit reports, eviction history, criminal background checks, income verification through bank or payroll connections, even detection of doctored documents and synthetic identities all run automatically. Used as document plumbing, this is genuinely useful: files get collected, verified, and summarized without a week of phone calls.

The line is decision-making. Scoring algorithms inherit whatever bias sits in their training data and present it back as a neutral-looking number; that was precisely the allegation in the SafeRent case, and fair-housing regulators on both sides of the Atlantic are paying attention. Data protection rules add their own constraints on what you may collect and how long you may keep it. Our recommendation is blunt: let AI collect documents, chase missing paperwork, and summarize files; do not let it decide who gets the apartment. The decision stays with you, with reasons you can state.

A related trap is geographic copy-paste. Screening features are deeply market-specific: a centralized eviction database exists in the U.S. but has no equivalent in many other countries, where screening rests on documents the applicant shares voluntarily. Translated listicles skip this difference; do not shop for features your market cannot deliver. We have written before about how algorithms are reshaping the property business from another angle, in our piece on what happens when the state runs the valuation algorithm.

What do the late-rent numbers say?

Late payment is the quiet enemy of rental income, and by recent U.S. data it is growing: the share of late rent payments rose from 8.8 percent in mid-2024 to 11.7 percent a year later. The U.S. consumer finance regulator CFPB adds a sobering detail from payment data: only about half of tenants who fall behind catch up the following month. A late payment is less often a one-off hiccup than the first sign of a trend.

What automation contributes here is not a miracle; it is discipline. Vendor blogs circulate figures like "automated reminders lift collections by X percent"; we could not verify any of them independently, so we will not repeat them. What we can verify is the mechanism: the reminder goes out before the due date, the payment is logged the day it arrives, and a delay shows up in your morning report before it is three days old. Manual tracking turns that loop into "I will check when I remember," and by the time the trend is noticed, three months have piled up.

A prerequisite hides underneath all of this: collections must flow through a traceable channel. Software can match a bank transaction to a lease; cash under the table produces no record to match, no history to score, and no evidence if a dispute ever lands in front of a judge.

A week in the life of an office managing twelve units

Let us make it concrete. A property office manages twelve rental units on behalf of their owners. In the old routine, the start of the month goes like this: bank statements are scanned to see who paid, non-payers get individual messages, two lease renewals turn out to have slipped past their dates, and one owner gets an awkward phone call because an increase was applied at the wrong rate.

Run the same office on proper software and the picture changes. Reminders go out automatically before the due date. Payments are logged; non-payers appear in the morning report. Upcoming renewals warn thirty days ahead, with the new rent precomputed from the current index. Applicant files (credit report, proof of income, references) are collected through the system, and missing documents are chased automatically. A few hours a week and several unpleasant phone calls a month come back.

The missed increase, in dollars

The cost of sloppy increase tracking is as concrete as multiplication. Take a unit renting at 1,500 dollars a month in a market where an 8 percent annual adjustment applies. Applied on time, the new rent is 1,620 dollars. If the renewal goes unnoticed for three months, the 120-dollar difference simply evaporates each month: 360 dollars gone, and the following year's increase now compounds from the lower base.

Now multiply by a portfolio. Across twelve leases with renewal dates scattered through the year, an office tracking by hand missing two or three renewals annually is not the exception. That is where entry-level software, often priced at a few dollars per unit per month, quietly pays for itself: one caught renewal covers years of subscription.

Frequently asked questions

Is rent collection software free?

Sometimes, at least for the landlord. Several U.S. platforms are free or nearly free on the landlord side and charge applicants for screening; general-purpose CRMs offer free tiers that small portfolios can adapt. Free tools rarely encode your local increase rules out of the box, though; you build that logic yourself, and your time is the hidden price.

What documents can I ask an applicant for?

The established set: a credit or rental history report the applicant shares with consent, proof of income, and a previous-landlord reference. The key principles are consent and proportionality: request documents from the applicant directly, collect the same set from every applicant, and do not retain the data after the decision. A standard set also protects you from claims of arbitrary treatment.

Can deposits and service charges be tracked too?

In good tools, yes. The deposit amount and its return conditions live on the lease record; service charges, repairs, and fixtures are logged as per-unit expenses. The year-end owner report then shows not just "how much rent came in" but the property's true net return.

What should you do?

• Start with an inventory. How many units, how many leases, which renewal dates? Even below five units, reminder and increase automation pays for itself; as the portfolio grows, software stops being optional.
• Wire the increase math into software, then verify the first cycle. Confirm the local cap and index rules are configured correctly by checking one or two renewals against the official published figures.
• Set a standard document set for screening. Ask every applicant for the same package: rental or credit report, proof of income, reference. Consistency is both fair and defensible.
• Never delegate the accept/reject decision to an algorithm. Let AI summarize, sort, and chase paperwork; make the final call yourself, with reasons. The SafeRent settlement shows the price of crossing that line.
• Watch arrears as a trend, not an event. React to two late payments in two months, not to one; if your reports cannot show that pattern, change your tool.

Choosing a tenant is a trust decision, and trust decisions demand reasons; algorithms tend to hide the reasons behind a score. Hand the machine everything routine (reminders, math, calendars, paperwork) and keep the signature for yourself. If you are weighing how to set this up for your own portfolio, we are around for that conversation.