Quantarium‘s founders, who have deeply technical backgrounds in computer science and quantitative genetics, weren’t seeking to disrupt a traditionally manual industry with artificial intelligence and machine learning – although that’s what happened.
Built on scalable cloud infrastructure, this company’s Automated Valuation Model (AVM) features deep learning algorithms that have processed data on over a 100 million property parcels in the U.S. The company works heavily with Computer Vision in particular, a field of AI that enables computers and systems to derive meaningful information from digital images, videos and other visual inputs — and take actions or make recommendations based on that information.
I sat down with Quantarium’s CEO and Co-founder, Clement Ifirm, to discuss the company’s AI-powered data-set, the future of the AVM and how Computer Vision has altered the scale and speed at which this data could be processed.
Romi Mahajan: Clement you founded a leading AI company in PropTech and residential real estate. As you think about the progress and evolution over the past years, how do you see Computer Vision as a key ingredient in real estate AI and data?
Clement Ifirm: When we think about Real Estate, the immediate impact of Computer Vision translates in the combination of Visual Sensors and Image Data Modeling – so Visual Hardware as well as Software, essentially an “Electronic Eye”, mixed with a “Visual Brain”.
As both hi-tech sectors progressed significantly over the past decade – while knowing that the capability to quantify property conditions at scale with a high level of precision is one of the most important challenges in this industry – it is clear that out of the larger AI/ML world, Computer Vision has a practical role now to significantly impact not only RE Valuations but the entire Underwriting process and the larger Real Estate and Mortgage space in general.
RM: Quantarium built a leading AVM, however, as off late AVMs have come under some attack. Can you comment on the power and usefulness of AVMs and how CV can enhance them?
CI: Certain events, mainly derived from the recent dynamics impacting the iBuyers space have raised some broad, though not yet focused, questions regarding the deployment of AI, the application of Automated Valuation Models (AVMs), and concerns for their precision and usefulness. Whenever a marque badge and venerable player elects to pull up stakes, abandoning a new high profile business model, while exiting a business line, it will clearly engender a lot of press attention and equal examination efforts by analysts who cover the space.
While the Quantarium team remains focused both on its customers and significant engineering progress in applying Computer Vision to AVM’s and the industry’s broad valuation challenges, we do pause long enough to look at this situation and encourage analysts and the press alike to sharpen their understanding for delineation among both the terms being bandied about and discerning root cause.
Not all AVMs are necessarily real AI models and manifestly an automated valuation model is not a business model – that later of course, comprises many factors, including critically, market conditions along with human judgement and execution.
While Quantarium, as any external entity, is not meaningfully privy to the defining data and the decision space confronting the management team, we would suggest it’s critically important to understand what failed here is the execution of a business model.
It becomes an arduous mountain top to summit, even with the best of climbing gear and plans.
Yet, other players in this space, ones who leverage similar tools and confront similar challenges, appear to be making progress against their forecasts which will yet again conclude a well known idealogy: business models often best succeed when technology is not postured as the sole answer, but rather deployed as a tool along with the best of human judgement to conquer new market opportunities.
We’ll leave the better forensic analysis and root cause determination to others, who have a grounded understanding of both the increasing capability of AVM’s and AI, along with the significant role and complexities of human judgement within the new market maker RE business model and what distinguishes the two.
Quantarium is seeing too much real, rapid progress for what a cost-effective AI based, Computer Vision assisted AVM can do for all industry stakeholders, to become preoccupied by a rash of snap, inflight analysis of certain AVM implementations within a very distinct business model.
Through capabilities such as image classification, categorization, grading and ranking, Computer Vision is continuing to increase the quality of RE Valuations and AVMs, with practical benefits to the RE/Lending space across both B2B and B2C. This is just the beginning, as next we will welcome further CV derived implementations such as point cloud solutions providing 3D reconstructions and AR technologies.
RM: So many PropTechs talk about “disrupting” the current ecosystem. Other authors, myself included, talk more about bringing cohesion into the marketplace while improving processes and experiences. What is your take on this?
CI: As we well know, while apparently antagonistic, both aspects are necessary elements: disruption and cohesion. Any vertical, including PropTech is subject to the laws of evolution, and looking through the lenses of genetics, while mutants are essential ingredients for progress, only some will “succeed” through certain variations of genes on a chromosome.
In other words, most entrepreneurial initiatives attempting to “disrupt” the PropTech space will just play their necessarily evolutionary roles to make the world a better place to live and shelter.
RM: Quantarium has some of the best AI and Computer Vision scientists on the team. You made that bet years ago, before CV was even a known quantity in the space. What was your vision at that time to make such bets?
CI: At Quantarium we are fortunate to be part of an exceptionally talented and experienced team, including math Olympiad champions, world class scientists and industry experts – all essentially defining the DNA of the company. Being High-tech at core, with a winning Research arm as part of its R&D group, Quantarium is constantly in touch with the latest technological advancements while still in their infancy, still in the labs.
Like Genetic Modeling and Evolutionary programming used in QVM, Computer Vision is one of the technological breakthroughs that Quantarium started experimenting early on, noticed its promising results and continued to invest in.
The Quantarium vision was anchored from the very beginning on finding and revealing the “Truth” by leveraging all possible pieces of information, all the possible data points. When we talk about Real Estate in general and RE Valuations in particular, there are two antagonistic dimensions to master: Precision and Coverage.
To paraphrase John Smintina, PhD, the Quantarium Co-founder we started the business with, the condition of a RE property is the “X Factor” in this industry and the Quantarium team, through the aforementioned research dynamics, recognized very early that Computer Vision was going to play a major role in mitigating and conquering this “RE X Factor”.
RM: What is the potential of CV in the residential real estate ecosystem / transaction ecosystem?
CI: Computer Vision technologies such as property interior and exterior feature detection, classification, condition rating and image compliance will continue to evolve and improve.
We may have heard about the “Cambrian Era” and “The Cambrian Explosion” which was an event that happened approximately 500 million years ago in the Cambrian Period when, during the span of ~ 100 million years, most major animal forms appeared in the fossil records.
Now, we can say that humanity entered the “Cambrian era of Robotics” and that is impacting life as we know it, including Real Estate.
The field of Robotics is improving rapidly, making devices, and building technologies by essentially mimicking natural selection. Keep what works, throw out what doesn’t, to optimally adapt a technology to a particular job.
At Quantarium, we are researching, building and deploying solution relying on both Machine Learning capabilities, learning from existing RE data, generating new, proprietary ML, Computer Vision based data, while experimenting with LIDAR and other Modern TOF (Time of Flight) Sensors to address the needs across the entire Real Estate Exchange pipeline, a true B2B2C play.
What’s next on the Computer Vision front, impacting all the dimensions of human lives including Real Estate: 3D capabilities such as mixed reality, augmented reality and virtual reality, a new world we have to understand and learn to live with and in.
RM: What is Quantarium’s future?
CI: Simply put, the Quantarium’s Vision & Investment Thesis translates into the fact that the Residential Real Estate space is on the cusp of digital transformation over the next years, as the Value Chain is impacted by AI/ML and broader data availability.
While the Real Estate vertical constitutes the largest asset class in the world, it is probably the last one to be significantly impacted by modern technologies. A few examples include regulatory broad acceptance to provide more waivers allowing AVMs & Hybrid Appraisals, as well as democratizing the data, allowing the Home Seller participation with impact into valuations for sale and refinance.
Quantarium, through transparent Artificial Intelligence that humans can understand and work with is architecting and delivering platforms and building blocks towards connecting the dots throughout the Residential Real Estate ecosystem, mastering the guts of the transaction and all the way to fulfilment, ultimately being the entrenched handmaiden to that approaching vector.
Talking about the future, Quantarium is advantageously positioned to play a special role in the up-and-coming Metaverse space – but that’s a conversation for another day.