What Happens When An Advanced Leasing Agent Becomes Free

As software cuts the number of people leasing premises, training will be necessary to do more advanced work.

Multifamily rental marketplace Apartment List announced that it would provide its advanced GenAI leasing agent Lea Pro to property management partners at no cost.

The move has some potentially disruptive implications. If it works as described, that could mean less need for employees to perform leasing activities. A cost savings, but also a loss of training opportunity for inexperienced people to learn the business.

The company describes Lea Pro as an “AI sales consultant,” capable of communications through email, text, chat, or phone. Apartment List claims that the software “nurtures leads and handles objections like a top salesperson, with smarter, scalable outreach” and “handles objections and knows how to persuade,” reading “emotional cues for more human-like conversations” with support in English and Spanish.

Although described as a generative AI system, the company has said that “Lea Pro has led over 7 million conversations and keeps getting smarter every day.” That would suggest at least some degree of machine learning in addition to a gen AI communications interface.

Apartment List claims that the software supports more than four million apartment units, has 68% more lead-to-tour conversions (although what the comparison is to isn’t clear), and saves 42 hours of work per property per week. Also, Lea Pro is available 24 hours a day. The focus is on contacting leads so employees “can focus on residents.”

“In today’s competitive rental market, leasing teams face unprecedented challenges, including record numbers of new apartments coming online and increasing vacancy rates,” the company said. “Lea Pro is designed to augment leasing efforts by being available 24/7 to answer questions and schedule tours, effectively moving prospective renters through the decision-making process.”

Two major questions arise. One comes from the technical end. An ongoing problem with generative AI systems is the inclination for them to “hallucinate,” making up answers, supposed facts, and references. This isn’t some bug that can be fixed so much as a feature of the unbridled use. Generative AI creates complex statistical representations of, within the bounds of the materials used in its training, how words commonly follow one another in particular contexts. But it doesn’t reason or store facts. The software can’t typically tell if something makes logical sense, as they aren’t designed to be logic engines. There are ways around the problem. BrainBox AI uses a generative query assistant to help building engineers. However, the software is restricted to using data only from the BrainBox system, effectively its ability to be “creative” in its answers.

It’s unclear how a system like Lea Pro would act with such restrictions, as persuasion, not just data delivery, is a goal.

A more subtle but concerning issue is how to train future property management personnel as software is designed, directed, and intended to as much as possible keep them out of an entire aspect of the business. Someone might have to step into a leasing process, and that means a need to gain experience over time to learn.