Methods of predicting a tenant’s risk of defaulting on a lease haven’t changed much over time. Generally, landlords look at things like credit score, generic financial data, and rent-to-income and debt-to-income ratios to make a final leasing decision. But, that’s all about to change.

Richardson-based RealPage announced Wednesday that it’s rolling out “RealPage AI Screening,” a machine learning platform that’s said to be able to predict a tenant’s risk of nonpayment, specifically in the multifamily apartment rental industry.

The technology not only assesses a tenant’s capability to pay rent, but also his or her willingness to do so. RealPage says by combining the two factors—and comparing them across millions of data sets—a better risk assessment model is created, which can predict overall financial performance.

According to a report in Property Management Insider, the AI taps financial and rental history records to estimate a person’s overall willingness to pay rent. Interestingly, the tech doesn’t consider rent-to-income ratios (these are said to be bad metrics that make screening less accurate), but looks instead at rent payment histories.

For example: Some renters have a history of consistently paying rent, even when money is tight and credit card bills go unpaid, the outlet reports. In the case of these renters, the AI model would prioritize their payment histories over their debt-to-income ratios and highlight them as reliable renters, even though a traditional model would most likely decline their applications.

Data is the fuel that drives artificial intelligence’s predictions, and RealPage has a hefty database from which to draw. The real estate data analytics giant says its prediction software analyzes trends in a database of more than 30 million actual lease outcomes, along with third-party consumer financial data and current market trends.

“We are confident in our screening model’s capabilities because it was tested and piloted by several industry leaders over the past six months, spanning more than 100,000 apartments,” Matt Davis, senior vice president of Financial Services at RealPage, said in a statement.

RealPage said the technology is more accurate than traditional screening models, which saves landlords and property managers an estimated $31 per apartment per year. This allows for the potential to return hundreds of millions in losses over time.

“Going from a statistical screening model to AI Screening is like jumping from an old rusty bike to a sleek, fast motorcycle,” James Flick, Camden’s director of revenue management, said in a statement to RealPage. “AI Screening looks for renters who don’t leave bad debt and lets the data tell us which renters are the best prospects for Camden.”