Where is the next up-and-coming area in a city? That is a multi-billion-dollar question for the real estate industry. Cities evolve over time, but their neighborhoods don’t change uniformly. Some neighborhoods gentrify and become more desirable, while others may face negative effects of growth, such as increased traffic or crime. The overall trajectory of a neighborhood might be intuitively felt by locals, but it remains one of the hardest aspects to grasp for outsiders.
This is precisely the challenge real estate teams face. Acquisition analysts must build an investment case for any opportunity, which means understanding how a neighborhood is changing. They often rely on real estate and demographic data to assess growth potential, but that only tells part of the story.
To truly understand where a neighborhood is headed, a broader perspective is needed. Teams must evaluate the local economy—what companies are employing people and how they are performing financially. They also need to consider local politics—what types of new buildings are permitted? Also, understanding local sentiment is key—what do residents think about the area? This comprehensive approach can be a monumental task, particularly for teams assessing numerous properties across a wide geography.
“I realized as a development consultant that there were often two very different viewpoints in the vetting process: the architect’s and the real estate team’s. Neither had the full picture of what was actually happening on the ground,” said Steven Song. Recognizing the need for a more uniform due diligence process in site selection, Song decided to develop a software program to collect all relevant information and even generate a score predicting a project’s potential outcome. He later co-founded Diald.AI to bring this tool to the entire industry.
Advances in AI have made it possible to digitize what was once an unstructured research process. “We start by casting the widest net possible,” said Song. “To do that, we scrape the entire internet for every vetted piece of information about the area.” This goes beyond a simple Google search for several reasons. AI can identify aliases for neighborhoods, which are sometimes referred to by the names of larger adjacent areas or have names that evolve over time or vary among locals. Also, AI doesn’t just locate reports and articles—it organizes and summarizes them in a structured way.
This broad data collection includes not only real estate and demographic data but also critical factors like local industries, development patterns, safety statistics, types of businesses, transportation options, zoning laws, infrastructure age, and nearby recreation. Once all the information about a property is gathered, the program assigns a score to the site.
The scoring system is designed to estimate the probability of a successful outcome. “We trained the model to analyze data from every past acquisition in the country and assess whether an opportunity shares characteristics with projects that defaulted or nearly defaulted on loans,” Song explained. A low score indicates a site has too much in common with failed projects. The extensive dataset Diald.AI uses enables it to weigh the significance of each similarity, helping establish causation rather than mere correlation.
Song acknowledges there’s no substitute for the instinct of a trained professional, but his tool equips experts with comprehensive information to make better decisions. “Early in my career, I worked on the development of the Revel Casino in Atlantic City, which famously turned into a disaster,” Song recalled. “I wondered how so many smart people could have been wrong about that decision, and I realized many of them simply didn’t have the site information needed to predict its reception.” AI’s ability to standardize data collection and analysis helps reduce potential biases within teams, while its speed allows for faster vetting of more deals.
Neighborhoods are not only defined by their buildings but also by the people who live, work, and play there. Researching a neighborhood, therefore, is as much about understanding local culture and customs as it is about analyzing the property market. Until now, the human-behavioral aspects of site selection have largely been an art. Thanks to Steven Song and Diald.AI, these qualitative elements are becoming more of a science, offering the potential for more predictable outcomes in large, high-risk real estate decisions.