Pegasus puts commercial leasing data toward AI joint venture

"Fore Real focuses on four services: property tax appeals, insurance compliance, lease abstraction and tenant communications. Pegasus provided the use of data from its portfolio as a participant in the joint venture. Fore brought technical expertise, accumulated over its 4-year history as a private company."

"By creating a product trained with a high volume of data points and then reviewed by a human, Jack Schoenfeld, Fore's chief growth officer and head of Fore Real, said the team was able to achieve high model accuracy, but also high human certainty. We embedded ourselves into Pegasus and did a lot of case work and a lot of prompt tuning and training on, literally, their brain, Schoenfeld said."

"They have over 1,000 tenants and so there's so many nuances and use cases and so that forced us to uncover [a variety of] edge cases where the logic is a little different. As the emergence of AI gets harder and harder to ignore, Pegasus CEO David Chasin said the firm had experimented with several other AI-driven technologies and automation tools but took issue with the levels of accuracy they were able to produce."

"The Fore team claims its models have 98 to 99 percent accuracy and in putting it to the test with his business, Chasin agreed, saying their work is the closest to perfec"

Fore Real is a partnership between an artificial intelligence firm and a commercial real estate advisory platform in Los Angeles. The offering streamlines tedious and costly tasks across four services: property tax appeals, insurance compliance, lease abstraction, and tenant communications. Fore Enterprise provides technical expertise developed over four years, while Pegasus contributes data from its portfolio, including assets under management totaling $4 billion. Fore Real uses high-volume data points to train AI models and then applies human review to validate outputs. The approach aims for high model accuracy alongside high human certainty, supported by extensive case work, prompt tuning, and training using Pegasus’s tenant and property nuances across many states. Reported accuracy is 98 to 99 percent.

Read at therealdeal.com