Errol Samuelson: Zillow Retools For Ai-driven Home Search — And The Agent Stays Central

Artificial intelligence is not only changing the way people search for and find information online, but it is also changing how they are searching for products and services, including their new home. Founded in 2006, Zillow has witnessed the housing industry develop from its early embrace of the internet to now entering the uncharted territory of an AI-first ecosystem. 

HousingWire recently caught up with Zillow’s chief industry development officer Errol Samuelson, who has been with the firm since 2014, to discuss how Zillow is adapting to the emerging AI-first world. 

This interview has been edited for brevity and clarity.

Brooklee Han: It seems like more and more people are turning to LLMs like ChatGPT for their search queries these days. How is Zillow adapting to the new ways consumers are searching for information and services? 

Errol Samuelson: What we are seeing broadly in technology now is that, whether you are doing a Google search or going directly to an LLM interface, consumers are now asking free-form questions. They are not dropping in keywords and then looking to receive 25 blue links as search results.

We are seeing this with people becoming more accustomed to [phrasing search] like, ‘I need a home with an office space because I work from home a few days a week, but also within 30 minutes of my office because I still go in a few days.’ Those are the kinds of conversations that, in the past, they would have with a human are now coming before they get to the city name, bed, bath and price part of the search. 

One of the things we are doing is enabling consumers to talk to the AI to figure out what they can afford and then using that, along with other [personal] factors, we can show them homes that match their lifestyle and budget needs. 

This isn’t just slapping an LLM on top of a listings database. To make the AI component more valuable, we are making sure that it has access to contextual information, like conversations the consumer is having with a Zillow-affiliated agent, allowing the AI to then suggest properties that are relevant to that consumer. The more context you can provide, the more personal and helpful the AI becomes. 

The other thing about Zillow’s AI that is unique is that the AI connects to the agent tools we supply. If you were to use a horizontal AI that does web crawling, it may find you properties for sale that meet your needs, but if you were then to ask if you can tour one of them, that is where a horizontal AI hits a wall and may suggest you call a local agent.

Because we have backend software components with our AI, we can actually use agentic AI to then go take action. So, if you tell it that you would like to see the property, it connects with our ShowingTime platform, and it can tell you when the property is available to be toured.

Additionally, since it also has that context, it can tell you if other properties in that same neighborhood are also available for tours around the same time or even if they have an open house scheduled. 

BH: As AI becomes more prevalent in home search, some organizations and individuals have raised concerns about potential fair housing issues. How is Zillow working to avoid any of these issues? 

Samuelson: Guardrails are super important. We put a lot of work into making sure that the responses and interactions that happen with the AI are appropriate by fair housing standards. We also released an open source fair housing model that the industry can use to ensure that conversations with AI are appropriate, but we do feel that there are some areas where advice is better provided by a human than by a machine. When our AI runs across these [situations] it will give consumers some things to consider, but tell them that they need to talk to a person about this. 

BH: There are constantly stories popping up about consumers using AI to sell or buy a home. What are your thoughts on this conversation that AI is going to replace real estate agents? 

Samuelson: There is this trope that AI is basically going to replace almost any profession, but what I think the AI actually does is it takes care of a lot of the low value work. It also helps the consumer be better informed and prepared for when they do work with the agent. Then, the agent gets to do the things they are good at that the AI can’t do like, judgement, negotiations, creativity and offering empathy and support. Buying a house is stressful, but by using AI, the agent is more available to help consumers navigate those very human emotions. 

Over the past five years, as this technology has emerged, we’ve seen that more consumers on both sides of the transaction use real estate agents. So, as we have seen more technology, the demand for the human in the loop is actually increasing, not decreasing. 

BH: There are so many AI tools popping up on the market. How do you feel Zillow’s longevity and experience in the industry will help it navigate this highly competitive and ever-evolving space?

Samuelson: There are definitely a few factors here, but one of the biggest is that we have been working with AI for a long time. Eight years ago, the Zestimate was incorporating neural networks and AI models, and we’ve had machine learning for a long time. We immediately doubled that work when we started some of these new models coming out back in 2017 and 2018. So, we have a lot of embedded in-house experience with AI, which is an advantage for us. 

The other thing is that we didn’t simply take an LLM or an AI model and slap it on our existing product. Instead, we tried to do exactly what we did with mobile, when we realized that mobile was going to be the way people use computing in the future. So, we completely turned the company on its head and went mobile-first. We are doing that now with AI and approaching things with an AI-first mindset and then looking at what that implies for the functionality that we want to provide to consumers and agents in the future. 

BH: As you just mentioned, Zillow has been around for a lot of technology innovation and integration in real estate, experiencing and learning many lessons over the firm’s 20 year history. What are some of those lessons the firm has learned in the past that you are working to apply to this latest technology evolution? 

Samuelson: The company was founded by the same people who created Expedia. They were looking to find ways to make information that was hard to access more accessible to consumers. With real estate, you couldn’t easily find out what your neighbor’s house had sold for, or what the approximate value of your home was, so the company was founded on this idea of transparency and that is still a core part of our ethos.

We have better transparency in the U.S. in real estate than anywhere else in the world and it makes for a more liquid market, which means it is easier to sell and price your home. That liquidity has led to a higher rate of homeownership in the U.S. and homeownership leads to building wealth, so a lot of good things come from transparency. We are deeply committed to maintaining transparency, cooperation and a level playing field for everyone in the country.

There are those in the industry right now who would like to turn back the clock and hide listings, price information, addresses, etc., and I think that is dangerous. Based on my time at Zillow, I firmly believe that we need to ensure that there is information transparency, so that is one big lesson. 

The second is that you have to start with the consumer and work backwards because it leads to better outcomes for everyone involved. Before mobile, before the internet, good agents have always worked hard to understand what their consumer needs and then provide that for them.

I think there is a tendency to create products or experiences from the inside out — working from the goal of closing a deal or getting another listing — and that doesn’t always result in as good of an outcome as starting from the outside and working in to the agent or broker. 

Finally, I think our industry tends to be a bit ‘wait-and-see’ when it comes to technology. We saw that with the advent of the internet and putting listings online, and we are seeing that now with [the] fear around AI. What we’ve seen with the internet, then mobile and now AI, is that each phase has an increasingly faster adoption curve. I don’t think the wait-and-see strategy is going to work with AI.

Zillow has a history of embracing technology and I think that lack of fear in adopting technology appropriately is definitely going to help us.