Artificial intelligence (AI) is likely to change how real estate and every other function in society is going to operate. The question is not “if” but “when,” and the answer, whether or not you are ready for it, is “right now.”
As new AI tools and use cases enter the real estate industry and, particularly, as they cross paths with data standards, RESO is dedicated to highlighting AI usage updates, no matter how quickly they are coming. And the pace has been blistering.
At the RESO 2024 Spring Conference, we had three AI sessions:
1) “Do LLMs Make RESO Obsolete?” presented by Justin Lundy, Founder & CEO at Lundy. (Hint: They do not.)
2) “Harnessing the Power of AI and Machine Learning with RESO Standards” presented by Lauren Martin, Senior Account Director at RentSpree; Santanu Barua, CTO at VestaPlus; Mathew Kallumadil, VP, Technology & Innovation at Stellar MLS; and Michael Lucarelli, CEO & Cofounder at RentSpree.
3) “Practical AI,” a roundtable moderated by Dave Conroy, formerly of the National Association of REALTORS® and now CTO of the California Association of REALTORS®, Greg Sax of RESO and Justin Lundy.
Do LLMs Make RESO Obsolete?
In the first presentation, Lundy delved into how Large Language Models (LLMs) and RESO standards work together to enhance real estate data processing and accessibility in unison. This culminated in a call for continued innovation and collaboration to bring more richness and accuracy to listing data. | WATCH VIDEO (12:45)
Real estate data is vast and generally messy, often overwhelming traditional processing methods. LLMs excel at interpreting unstructured data, finding patterns and generating insights. However, for LLMs to function effectively in real-world applications, clean and standardized data is crucial.
RESO standards provide a structured format for data exchange, ensuring consistency and accuracy, and allowing LLMs to focus on their strengths – analysis and interpretation – rather than data cleaning.
Lundy, the company, utilizes LLMs to create a natural language interface for real estate data, but they encountered hurdles when integrating LLMs with their data platform. Their initial excitement about the “jaw-dropping” capabilities of LLMs in a prototype turned into concerns about processing times and costs when dealing with real-world variations in data formats from different sources.
Justin Lundy, Founder & CEO of Lundy.
This problem was solved by RESO standards. Standardized data meant less development time, faster processing speeds and less financial burden.
Thinking about cost per user isn’t something that a motivated entrepreneur wants to dwell on, but Lundy had to heavily weigh every single interaction until using RESO-certified metadata sped up their onboarding. Even partially compliant data significantly improved LLM performance and reduced development costs.
Upon this revelation, Lundy rewrote 75% of their phrase-based code to better understand the RESO-standardized data sitting in MLS systems, and it was, according to Lundy, an incredible leap forward.
Further emphasizing the role of RESO, Lundy said, “You have been given access to billions of dollars worth of AI that you can harness to shape this industry into something we can’t even imagine, but we can’t do that if all of our money and time goes into processing messy data.”
Lundy envisions a future where standardized data and AI tools work in tandem. By leveraging the strengths of both RESO standards and LLMs, the real estate industry can unlock a new level of data accessibility and empower users with powerful insights regardless of their background or technical expertise.
Harnessing the Power of AI and Machine Learning with RESO Standards
The second AI presentation was a panel discussion that explored how real estate professionals can leverage AI and machine learning alongside RESO standards to unlock new possibilities in data analysis and user experience. The panel featured expert perspectives from MLS and proptech representatives. | WATCH VIDEO (19:08)
Barua from VestaPlus noted how AI can help with listing input beyond required fields, improving data quality by filling in missing fields in listings based on image recognition and descriptions, reducing human error and manual entry.
Barua also talked about how AI is enhancing not just listing inputs but search functionality with plain English or audio prompts.
According to Lucarelli from RentSpree, a more personalized approach to search extends to consumer recommendations as well. Based on a case study in collaboration with CoreLogic using a cohort of renters from two years ago, RentSpree was able to harness AI to analyze renter data like income, credit score and ZIP Code to predict purchase intent.
From there, a feature could be created to recommend properties with a higher likelihood of conversion, whether individuals are looking to rent or buy. As a consumer, imagine being presented with a set of listings with more chance for approval, eliminating the frustration of applying to multiple properties that are ultimately out of reach.
Left to right: Lauren Martin, RentSpree; Santanu Barua, VestaPlus; Michael Lucarelli, RentSpree; and Mathew Kallumadil, Stellar MLS.
Barua also noted how AI can be used to automate internal MLS tasks that were previously time-consuming for staff.
For example, it can identify copyright violations in listing images, freeing up human resources to focus on more complex issues. This automation extends to document processing as well. It can also convert unstructured listing agreements and seller instructions into structured data, streamlining the workflow and eliminating the need for manual data entry by compliance staff.
Kallumadil from Stellar MLS talked about how they are using AI for education and support. Not only are they training staff on prompt engineering, but they have uploaded numerous transcripts into their system to serve as a support Q&A. Now, newly onboarded staff can find what they need more easily.
The goal for Stellar MLS is to keep training the AI to serve the consumer directly on its own in what has become the ubiquitous chatbot. This is something that RentSpree has achieved and continues to train.
The panel also pondered how AI might evolve to automatically interpret and utilize data beyond the current RESO structure, potentially leading to adjustments in the RESO Data Dictionary.
Standards and AI Clearly Go Hand in Hand
All of the panelists agreed that clean and standardized data is crucial for training effective AI models, as inconsistent data formats create hurdles and slow down development. RESO standards also ensure data consistency and facilitate a smooth workflow for AI integration.
The experiences of our conference speakers serve as a powerful testament to the enduring importance of RESO standards in the LLM/AI era. RESO functions as an essential tool for enhancing LLM effectiveness, and the future of real estate data involves continued collaboration between standardization and powerful AI.
Practical AI Beyond RESO Standards
Dipping into waters beyond data standards, the third of the three AI conference sessions was a roundtable that delved into practical applications of AI in day-to-day work.
Justin Lundy and Dave Conroy at the “Practical AI” roundtable.
A spirited and open discussion occurred about how people are currently using AI. The gathering was most interested in the legal ramifications of AI, especially copyright, as vast libraries of content are constantly being opened up to more AI tools.
The tools that were discussed beyond the much ballyhooed ChatGPT include Google Gemini, Microsoft Copilot, Meta AI, Midjourney, Ideogram and more.
There are already countless hours of video and written tutorials about how to maximize AI functionality with instructional prompts that can help you with email management, calendar organization, coding, creating images, starting businesses and, yes, writing blog posts such as this one. Learning how to use AI was not the purpose of the roundtable discussion.
The roundtable provided a firm shift in tone to the fact that bleeding-edge real estate professionals are already using AI today. It was reminiscent of discussions about how first adopters were using the Internet to enhance the real estate industry at the dawn of the 21st century.
Every new technology cycle comes with a learning curve. We encourage you to embrace the change.
