Domain Expertise Still Trumps AI in Real Estate Data

Written by:

Rudi Davis

Date:

10 Feb 2026

Share

Share via Facebook

Share via Linkedin

Share via X

Copy to clipboard

As artificial intelligence and web scraping tools become more accessible, many assume that large technology companies with advanced AI capabilities will dominate the structuring of real estate data. Daniel Heller, CEO and co-founder of ReZone (recently acquired by Shovels AI), disagrees. She argues that the absolute advantage in real estate data lies not in computational resources, but in domain expertise—knowing which data is relevant and which is not.

“There is a lot of data out there that is just complete garbage,” Heller says. Years spent building a company that tracked city council and planning department decisions nationwide taught her that insight into what matters is more important than access to vast data sets. This perspective challenges the idea that general-purpose AI tools will inevitably control specialized data markets.

Domain Knowledge as a Differentiator

Heller sees real estate as a sector that large technology companies routinely overlook. “Real estate being one of them, where people don’t get enough love from big companies,” she says. This lack of attention leaves space for companies that understand the specific, often complex questions real estate professionals need answered.

She distinguishes between broad, generic questions and the specialized inquiries that drive real estate decisions. For someone asking fundamental questions about investing in real estate, she acknowledges that general AI tools like ChatGPT will suffice. But when the question becomes more specific — such as identifying which developers are active in a particular market — publicly available information is sparse, and specialized knowledge becomes essential. “There is so little information to be found in that regard that there’s a lot of opportunity,” Heller explains.

As AI improves at answering general queries, Heller believes the value will increasingly shift to companies capable of addressing particular, domain-dependent questions. These questions require not just data, but judgment about which sources are trustworthy and which are simply noise.

The Reality of Data Abundance

Heller argues that the proliferation of AI tools and inexpensive web scraping does not eliminate the need for specialized data companies; it may actually heighten it. The ability to collect and process vast amounts of data is now widely available, but that does not guarantee better results. “It will probably make some database companies or data companies a little bit useless,” she says. “But it will also create the chance for companies to really excel and do this work for you, and provide you with information that is the most valuable; it can be updated every minute.”

She points out that the declining cost of processing data, “the cost of a token to be non-existent,” does not automatically yield higher-quality insights. Instead, it raises the stakes for companies to distinguish relevant information from irrelevant or misleading data. The companies that have invested in understanding their markets deeply enough to make these distinctions accurately will be better positioned than those that run AI models on large, undifferentiated datasets.

Why Specialized Companies Survive

Heller’s experience with ReZone demonstrates that niche data companies can thrive even when larger players have theoretical advantages. ReZone monitored every city council and planning department meeting nationwide, focusing on commercial real estate news such as zoning decisions, economic development proposals, and permit approvals, details that are often buried in local government records and difficult for outsiders to interpret.

The company’s acquisition by Shovels AI, which specializes in permit data, was based on the belief that combining these different specialized data sources would yield insights that general-purpose tools cannot easily replicate. “We believe that combining city council data with permit data would create a layer of insight that was not possible today,” Heller says.

This strategy, integrating multiple specialized datasets with a deep understanding of what matters in real estate, offers a more sustainable advantage than competing with general AI tools on sheer data volume.

The Next Phase of Data Competition

Heller’s observations highlight a changing landscape in real estate data. While some database companies may lose relevance as scraping and processing become cheaper, others will succeed by focusing on curation, context, and domain-specific structuring that general AI still struggles to match.

“There are a lot of data sources that are unstructured enough that we haven’t actually looked into them yet, but now that we can structure them, the cost of a token is non-existent,” Heller says. She sees an opportunity for companies that can identify which unstructured sources contain relevant signals for specific real estate use cases and possess the expertise to structure and interpret them accurately.

The speed at which specialized companies can claim these niches may determine whether they remain ahead of general-purpose AI tools. For now, Heller’s experience suggests that domain expertise remains the key differentiator in making real estate data truly valuable, especially as the volume of available information grows and the challenge shifts from collecting data to knowing what to do with it.