The real estate industry is entering a new era where artificial intelligence is reshaping how properties are evaluated, monitored, and managed. From AI-powered image recognition to predictive analytics, emerging technologies are helping lenders, insurers, and investors gain deeper insights into property condition and collateral risk with greater speed and consistency than traditional assessment methods.

In this interview with AI Spectrum, Eric Fox, Chief Economist & Senior Vice President, Analytics at Veros, explains how VeroVISION is advancing AI-driven property intelligence through computer vision and automated condition scoring. He discusses the growing role of AI in enhancing AVMs, streamlining risk assessment, and enabling smarter decision-making across the broader real estate ecosystem.

VeroVISION uses AI-driven image recognition to evaluate property and room conditions. What technological advancements made it possible to achieve a 93 per cent correlation with human appraisers?  

The key technological advancement that has allowed this level of accuracy has been the recent advances in computing power.  In particular, GPUs allow the rapid processing, identification, and scoring of millions of new photos every day.  To date, Veros has processed over a billion photos within VeroVISION.   

How do you see AI-powered property condition analysis transforming the future of AVMs and real estate risk assessment?  

This will only make AVMs even more accurate.  In the past, one limitation of AVMs was that they could not assess the condition of the property.  Now with VeroVISION, property condition can be accurately assessed.  This means that instead of selecting comparable properties simply based on nearby properties with similar property characteristics (such as square footage or number of bedrooms), comparables with similar conditions to the subject can likewise be selected.  Also, properties with significantly better or worse conditions than are typical for the neighbourhood can have their values adjusted upwards or downwards, respectively. 

What are the biggest challenges in driving industry trust and adoption for AI-based property analytics over traditional manual inspections?  

Just like the adoption curve which occurred for Automated Valuation Models (AVMs), there will be a period of validation and testing which will need to happen for condition scoring.  However, based on the results of 93% correlation between the Veros Home Score and human appraiser score on the same photos, the accuracy is already quite advanced and is a significant development for real estate valuation.  There will always be a need for some form of manual inspection for some properties – especially for complex, non-conforming properties.  But AI opens the possibilities of what could happen in the future.  For example, the homeowner could take a video of the property under the direction of a remote inspector, and the resulting video could then be scored by AI to accurately assess the condition.  Certain situations will still dictate that a traditional manual inspection must occur.  AI-based property analytics simply give the industry more options for assessing the property’s condition in a cost-effective manner.

How does the Veros Home Score enhance decision-making for lenders, insurers, appraisers, and institutional investors?  

The Veros Home Score only enhances the decision-making process by providing “another set of eyes” on the property and its likely condition.  For example, many appraisals are currently “auto-populated” with a condition of “average.”  The Veros Home Score could flag properties which are likely to be in poorer condition than “average” for enhanced processing.  This would lead to a reduction in overall collateral risk. 

What role will computer vision and predictive analytics play in improving operational efficiency and reducing collateral risk as AI adoption accelerates?  

Assessing collateral risk is not a one-size-fits-all.  Computer Vision and predictive analytics allow just the correct solution to be used for the specific property being assessed.  Real estate valuation will become a continuum depending on the risk spectrum, ranging from lowest risk loans being assessed by a low-cost AVM all the way to the highest risk loans undergoing a high-cost full interior inspection and appraisal.  AI adoption and tools such as VeroVISION will accelerate options between these two extremes.

Do you see AI-enabled property intelligence evolving into broader applications such as asset management, insurance automation, or smart real estate ecosystems?  

Of course.  Portfolio asset management is the next natural step for AI.