Arx’s Approach to Underwriting Deals With AI

Tomas Garcia left his job as an investment banking analyst with Goldman Sachs to fix and flip homes with his brother, Hernan. Those experiences inspired them to create Arx, a platform that uses artificial intelligence (AI) to allow real estate investors and developers to accurately source, evaluate, and underwrite land and building projects in mere seconds instead of weeks, months, or years. 

By saving time in the underwriting process, Arx also speeds acquisition and development, an aspect that Garcia hopes will boost housing supply and ultimately help resolve the industry’s widespread affordability crisis. He calls Arx’s AI system a “force multiplier” that doesn’t seek to eliminate the jobs of underwriters, but rather to heighten their success rate, reduce costs and time, and increase the number and quality of the deals they close.

PRO BUILDER: What is Arx and what does it accomplish in today’s real estate market?

Tomas Garcia: Arx is an AI-driven real estate analytics platform that underwrites the future potential of millions of properties in advance. It empowers developers, investors, real estate agents, and the brokers who work with them to source and evaluate optimal investment and development opportunities in seconds. 

It was born out of the frustration my brother and I experienced with the incredibly inefficient process of deal sourcing, which results in a lot of time wasted. 

We knew there had to be a more efficient solution, so we set about building this massive artificial intelligence system that can actually underwrite the market in advance and allow developers and investors to identify sites based on financial characteristics, enabling them to quickly zero in on properties that have a high probability of becoming profitable deals. 

RELATED

• There’s a New Sheriff in Town at NAHB: Jim Tobin Becomes CEO
• John Green’s Plan to Eliminate Predatory Lending Practices
• Introducing Modular Construction to a New Labor Force

PB: What are the capabilities and bandwidths of the AI system you’re using? 

TG: We have a lot of different datasets flowing into the system, including tax assessment data, building footprint, construction cost estimation, building permit data, parameterized zoning code information, topography, and more—basically, as many different datasets as you can think of that affect real estate.

The system currently supports for-sale new residential construction and infill development, from single-family to quadruplexes. Right now, we’re operating in Los Angeles and Seattle, with a near-term goal of serving the entire U.S. We want Arx to cover larger-scale assets and different investment strategies, and we’re currently in the process of incorporating rentals as well as renovation projects into the system.

PB: How much research can clients do and how accurate is the information they’re getting? Can you guarantee that accuracy?

TG: Accuracy is very important to us, and we're trying to make recommendations to people but in the language that they understand. For the system, the output, and the financial model, all of the assumptions are that the data is pre-populated based on algorithms, but users can fine tune them.

Let's say the system predicts that it will take five months to get through designs and approvals in a particular municipality, but you have an expediter that can get you through in three months. You can make that adjustment and everything updates for better accuracy.

In some cases, it takes years to do what our system does in seconds, so in this instance, AI is making employees more efficient and allowing companies to grow instead of being stuck in a quagmire of lost deals and lost time.

The best sort of barometer that we currently have is the feedback we’ve gotten back from our users. We're hearing from clients that the system is underwriting projects to within about 5% on all the assumptions of how they would underwrite their own. We're trying to stay within that 5%.

PB: How does AI actually help people be more productive versus replacing them? 

TG: We view our tool as a force multiplier not as a replacement for individuals. Depending on the size of the shop, there could be anywhere from two to 40 people focused on the specific problem our tool is targeted to help solve. 

In talking with them, we found they’re wasting up to 90% of their time because many of the deals they’re working on get killed before they come to fruition. 

A lot of companies can’t afford to hire more people just to get that 10% of deals that actually work. With Arx, however, the same number of employees can look at a lot more deals. In some cases, it takes years to do what our system does in seconds, so in this instance, AI is making employees more efficient and allowing companies to grow instead of being stuck in a quagmire of lost deals and lost time. 

PB: Why is this software particularly useful in today’s market? What inefficiencies is it bypassing or simplifying? 

TG: One of the main challenges we face across the country is very complex regulation that affects what you can build and, specifically, the profitability of what you can build. Our system is taking those regional regulations into account, so, for example, when we’re simulating new developments, we’re simulating the impacts of a transfer tax on the actual profitability of the project. 

Arx ultimately bridges that fundamental disconnect between market forces that builders want to capitalize on and the limitations local policy and regulation dictate. At its core, our system is helping real estate professionals thread that needle to figure out not just what they can be doing in a particular market, but what they should be doing there, as well.