Zillow's Expanding AI Bet: What's Actually New and What It Means

Zillow has spent the past two years integrating AI more deeply into nearly every part of its platform, and the changes have been incremental enough that it's easy to miss how much has actually shifted. A closer look at what the company has built — and where it's heading — tells a more interesting story than any single announcement.

Natural Language Search

The most visible change for consumers is the move toward natural language search. Where users once filtered by bedrooms, price range, and zip code, Zillow now lets buyers describe what they're looking for in plain language: "a three-bedroom with a big backyard near good schools, walkable to coffee shops." The system interprets those inputs against its listing database and returns results that match the intent, not just the filters.

This is a meaningful shift in how search works on the platform. For agents, it means buyers are arriving at properties through a different kind of matching process — one that may surface listings they wouldn't have found through traditional filters, but also one that isn't always transparent about why a given property was recommended.

Zestimate Improvements

Zillow's Zestimate has long been both its most-used feature and its most-criticized product. The company has made several behind-the-scenes improvements to the underlying model, incorporating more granular data sources and updating valuations more frequently. The accuracy improvements are real, though Zestimate still struggles with properties that are unusual for their market — a consistent limitation of any automated valuation model.

More interesting is how Zillow is positioning Zestimate relative to its agent partners. The company has emphasized that Zestimates are a starting point for conversations, not a substitute for a comparative market analysis. Whether agents find that framing credible probably depends on how many of their listing conversations have started with "but Zillow says..."

The Strategic Picture

Zillow's AI investments make sense when viewed against its broader business model. The company earns most of its revenue by connecting buyers and sellers with agents through its Premier Agent program. Anything that makes the platform more useful — that keeps consumers on Zillow rather than searching elsewhere — protects that business. AI-powered search and better recommendations are fundamentally about engagement and retention.

For agents, the practical implications are straightforward: the platform buyers use most heavily is getting better at influencing which homes they see and in what order. That's worth understanding even if the underlying algorithms aren't fully transparent.

- Jason