A One-Click Real Estate Transaction?
It is Not Entirely Out of the Realm of Possibility
By Carter Pratt and Bryan Robinette
The integration of digital automation in the real estate industry has made valuation, diligence and title services far more streamlined than they had been in the analog days. Today, with artificial intelligence-powered tools coming into the market, the dream of near frictionless real estate transactions seems like an increasingly likely reality.
Real estate investing has always been a complicated business involving multiple stakeholders who participate in the transaction, from the buyers and sellers, lenders and lawyers, title companies and notary services all with their own stacks of paperwork to file, verify, sign, and register. In the not-so-distant past, the process of transferring property from one owner to another was an arduous, non-digital, labor-intensive operation of parties manually moving physical documents from one person to another with plenty of room for human errors, any one of which could have dire consequences for the entire transaction leaving all parties bewildered at the conclusion of the process.
The Advent of Digitization
The advent of digitization was a helpful step forward for both accuracy and speed. The ability to capture documents as digital images allowed paperwork to be filed, sorted, duplicated, and shared among all parties in a common format. For the title industry, uniform records helped eliminate some of the variability that had made searching for title in the past more of an art than a science. For other functions, the ability to see and compare documents helped investors make quick assessments of multiple properties.
In a sense, digitization helped change real estate investing from a paper-bound business to a data processing one where information from multiple sources can be extracted, transformed, and loaded into a useable data repository before strategic analysis.
Much of this transformation was jumpstarted because of the foreclosure meltdown and financial crisis that followed it in 2008. Deluged with available properties, investors struggled to evaluate opportunities in a heavily distressed market. The legwork required to execute an investment strategy was carried out with actual legs belonging to a national network of brokers, agents, and inspectors who could provide on-the-spot assessments of the condition of each property and an estimation of the cost of any repairs. Then the Covid-19 pandemic further complicated these already inefficient processes by making physical assessments nearly impossible just as the single-family rental (SFR) market was taking off.
The industry has solved many of these challenges with technology. Today, there is a digital ecosystem that enables investors to assess dozens of properties, compare one to another, repair what is necessary, and then sell them without ever having seen the property in the real world. This ecosystem is powered by an automated array of technology features such as optical character recognition (OCR), robotic process automation (RPA), and extract, transform, and load processes (ETL). Working in concert, these technologies help review the content of hundreds, or even thousands, of documents and identify anomalies for additional review.
A Decade of Progress
It is amazing to look back and see how far we have come in a short amount of time. Just a decade ago, many of these technologies that the real estate investment industry now depends on did not exist. For example, the due diligence process was nearly entirely manual a decade ago. Due diligence required highly trained professionals to physically review mortgage documents, title policies, lease agreements, and other recorded documents and manually type inputs into a system. They would spend hours doing side-by-side comparisons of documents and data values to make sure they matched—a tedious process appropriately nicknamed stare-and-compare. As with any manual process, this introduced issues. For example, inconsistencies in how each reviewer entered information oftentimes resulted in repeating these input tasks in order to standardize everything.
Today, by overlaying OCR and RPA technology, much of this work can be automated which can help to increase the ability to scale operations efficiently and accurately. This frees up the professionals to focus their attention on higher level decision-making rather than time-consuming and repeatable data entry.
However, there are still limitations with the current generation of technology. While optical character recognition is able to extract data from a document, it lacks the ability to decipher nuance or context. But with the introduction of generative AI and large language models (LLM), technology is now better able to contextually understand the content of those documents through question prompts. For example, rather than sifting through a 150-page PDF document, a title insurer could ask, “Who is on title for this property?” and have that information retrieved in a consistent and integrated way.
The potential of overlaying AI on top of automated processes could also help avoid some other common chores. AI can help us go beyond extracting data by interpreting the data based on nuances of the property location. For example, recording requirements vary from one of the 3,000+ counties in the United States to another. Generative AI trained on these requirements has the potential to help those without a career’s worth of experience instantly apply these requirements for recording a mortgage in Los Angeles or the requirements for clearing a lien in Westchester.
Looking Ahead
At the Radian family of companies, we are making huge strides towards judiciously exploring these capabilities. Over the past several years, in a controlled environment, we have invested heavily in building and training AI models to help improve, accelerate, and simplify transactions for clients.
Pre-AI enhancements to homegenius Real Estate’s Pyramid Platform operating system and Single-Family Rental Due Diligence and Valuations Dashboards provided by Radian Real Estate Management have helped investors enhance their automation in nearly every step of the funding process based on their internal requirements and the bank’s requirements. These platforms can be accessed by the investor, the capital markets lending institution, and the diligence reviewer so that all parties can see what is occurring with their transactions in real time.
A user can see at a glance the diligence requests for each property, who the lender is, what assets are funded, what documents have been reviewed, and what assets are not yet funded because documents are still pending review. In today’s operating environment, using an ETL process and overlaying OCR and RPA, the dashboard can also help speed up the BPO valuation and allows each user to configure the information according to their own preferences, which helps avoids compatibility obstacles and helps make the whole process more efficient for everyone.
Now comes generative AI. It has the potential to make this system even more seamless while adding agility to the processes. The promise of AI is that it can help make transactions frictionless and instantaneous from the point of origination.
In an ideal future state, generative AI could fully automate the property acquisition, funding, and potential securitization processes so that an investor would merely have to input their requirements and approve transactions as they go. Imagine getting an alert as soon as a property that fits the investment criteria is available. With the appropriate documents, valuations, funding, and approvals all automated, transactions could take place instantaneously at the click of a button. Suddenly, growth strategies can be scaled much more quickly.
Once a property is in the system, all of its data is accessible for future transactions as well. Suppose that same investor wants to pool their assets to get more funding. They would potentially be able to easily pool a dozen properties in Nevada, Utah, and Arizona to request a new funding line. The system could then perform an automated review and alert the appropriate party if any new documentation is required. All that is left is for the bank to approve and release the funds.
A few years down the line, perhaps that investor wants to do a securitization. This time they have not only pooled the properties in those three states but now they have added 4,000 properties across 25 states. The process remains the same, whether it is 40 assets or 4,000. With all the documents in one place and all the requirements known, the reviews complete, and inconsistencies resolved, an AI-powered system with these capabilities has the potential to make the whole process smooth and almost immediate. This automated efficiency is the potential that generative AI may have for investors.
How long until that vision becomes a reality? What sounds like a far-fetched fantasy may not be that distant after all. While the internet took about a decade to become widely adopted by the public and smartphones took about 5 years to become ubiquitous, generative AI technologies have seen substantial adoption in just the past year. For example, ChatGPT reached over 100 million users just two months after launching. Although not yet in production, testing and research continues as Radian attempts to realize the positive benefits of generative AI models for real estate transactions.
The power of generative AI is that it continues to evolve at a rapid pace. What we are creating today at Radian may very well be the foundation for the next phase of real estate investing tomorrow. Considering how far we have come is such a short time, a one-click transaction may not be entirely out of the realm of possibility.
Whether it will be our reality in the foreseeable future is something only time will tell.
