AI In Real Estate: Where To Start

Marc Rutzen serves as cofounder and CEO of HelloData.ai.

The world of real estate is experiencing exciting advancements thanks to AI, machine learning and computer vision. These technologies offer new ways to improve efficiency and tackle issues like automated data capture and fraud detection. In this article, I want to share some ideas about how you can start using AI in real estate (without getting too technical).

Handling documents in real estate can be time-consuming. Different formats and inconsistent templates make it challenging to extract and organize data. AI can help by automatically extracting relevant information from complex documents like leases, appraisals and loan documents.

With recent advancements in optical character recognition (OCR) technology, even poorly scanned or handwritten documents can be detected with high accuracy. This means important data like property values, loan terms and more can be easily identified with AI.

With some engineering, this data can be brought into existing workflows. For example, commercial mortgage lenders usually want to capture the values and cap rates in appraisals for underwriting and loan sizing. If AI is used to extract this data and something like Zapier is used to move it around, you could automatically extract data from appraisals uploaded to Box, SharePoint or something similar and push the data to Salesforce.

This would give everyone in the company access to valuable deal data with a simple search.

Writing listing descriptions is a tedious task for brokers, but it can be automated with AI. Although there are paid tools available, you can start for free using tools like ChatGPT. Simply create templates with property details and prompts and input them into ChatGPT, and you'll get a draft listing description in seconds.

However, it's important to double-check the generated descriptions for accuracy, as AI can sometimes make mistakes or make up information. Incorrect information in multiple listing service (MLS) listings can still lead to liability issues, even if an AI wrote it.

With AI-generated images and listing descriptions, there are new challenges for MLS providers. Manipulated images and deceptive descriptions can be produced much more easily and can just as easily go undetected. This is a major problem for MLS administrators and brokers alike, as there are many fines and penalties for inaccurate listing data.

AI can be used to address these issues, too. Computer vision and AI algorithms can identify listing inconsistencies by comparing listed features in listings with those in property images. They can also do things like detecting logos or signage from a competing brokerage in a listing, identify signs of image manipulation and even help flag potential Fair Housing Act compliance issues.

All of these things can help MLS administrators ensure accurate and compliant listings.

AI can assist in cross-checking and verifying data between documents, which is crucial during the loan application process. Algorithms can quickly compare data points across various documents to identify inconsistencies or inaccuracies. With all the manual data entry in real estate, there is a lot of room for error.

Key values from documents like lease agreements and property condition assessments can be extracted and compared to find issues like mismatched values or different addresses listed on each document. Inconsistencies can be flagged for follow-up, helping reduce the risk of fraud or delays in the loan approval process. These applications could be useful to both commercial mortgage lenders and organizations like Fannie Mae and Freddie Mac that purchase loans.

Even though the adoption of AI in the real estate sector offers incredible potential, there are some challenges. Data privacy is a big one, as AI applications often require access to personal and financial data, raising concerns over secure data storage and ethical use. Many lenders, brokers and appraisers I've spoken with are concerned about their work product being used to train generative algorithms that compete with them—and it’s not an unreasonable concern.

Another risk is bias in algorithms and the potential for fair housing violations. AI is proficient in identifying patterns, so it can make recommendations based on initial user preferences, then learn from their selections to continually narrow results in a housing search, for example. Depending on the attributes used and how narrow the search results become, this could effectively steer people to certain properties and neighborhoods. Steering is steering, even if an AI did it.

Finally, the adoption of AI technology could inadvertently widen the digital divide, making it more difficult for entry-level analysts to navigate real estate transactions. I’ve heard countless times how important it is for underwriters and analysts to “get in the weeds” and understand the process by doing the work—automating their work with AI could make this more difficult.

That said, overall, AI is revolutionizing the real estate industry by improving operational efficiency, reducing errors and automating manual tasks. While the examples mentioned here are just the beginning, they demonstrate the immense potential of AI in real estate. As we embrace these technologies further, we can expect even more transformative changes in the future.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?